diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4742 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 200,
+  "global_step": 668,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0029940119760479044,
+      "grad_norm": 20.101527338485713,
+      "learning_rate": 9.999944704978835e-06,
+      "loss": 1.2177,
+      "step": 1
+    },
+    {
+      "epoch": 0.005988023952095809,
+      "grad_norm": 19.79491053983428,
+      "learning_rate": 9.999778821138357e-06,
+      "loss": 1.2362,
+      "step": 2
+    },
+    {
+      "epoch": 0.008982035928143712,
+      "grad_norm": 4.180070127910943,
+      "learning_rate": 9.999502352147583e-06,
+      "loss": 0.775,
+      "step": 3
+    },
+    {
+      "epoch": 0.011976047904191617,
+      "grad_norm": 5.846095333645197,
+      "learning_rate": 9.999115304121459e-06,
+      "loss": 0.8757,
+      "step": 4
+    },
+    {
+      "epoch": 0.014970059880239521,
+      "grad_norm": 1.266570274313935,
+      "learning_rate": 9.998617685620715e-06,
+      "loss": 0.5859,
+      "step": 5
+    },
+    {
+      "epoch": 0.017964071856287425,
+      "grad_norm": 1.3925918631997296,
+      "learning_rate": 9.998009507651683e-06,
+      "loss": 0.6823,
+      "step": 6
+    },
+    {
+      "epoch": 0.020958083832335328,
+      "grad_norm": 1.4831281338169187,
+      "learning_rate": 9.997290783666048e-06,
+      "loss": 0.6537,
+      "step": 7
+    },
+    {
+      "epoch": 0.023952095808383235,
+      "grad_norm": 1.2114430633800772,
+      "learning_rate": 9.996461529560553e-06,
+      "loss": 0.5682,
+      "step": 8
+    },
+    {
+      "epoch": 0.02694610778443114,
+      "grad_norm": 1.0071699938425211,
+      "learning_rate": 9.995521763676645e-06,
+      "loss": 0.6282,
+      "step": 9
+    },
+    {
+      "epoch": 0.029940119760479042,
+      "grad_norm": 0.9045658366046047,
+      "learning_rate": 9.994471506800078e-06,
+      "loss": 0.5864,
+      "step": 10
+    },
+    {
+      "epoch": 0.03293413173652695,
+      "grad_norm": 0.8506052536688817,
+      "learning_rate": 9.993310782160439e-06,
+      "loss": 0.6068,
+      "step": 11
+    },
+    {
+      "epoch": 0.03592814371257485,
+      "grad_norm": 0.8037108557555585,
+      "learning_rate": 9.992039615430648e-06,
+      "loss": 0.5642,
+      "step": 12
+    },
+    {
+      "epoch": 0.038922155688622756,
+      "grad_norm": 0.7267419767340054,
+      "learning_rate": 9.99065803472638e-06,
+      "loss": 0.5091,
+      "step": 13
+    },
+    {
+      "epoch": 0.041916167664670656,
+      "grad_norm": 0.6712047651531428,
+      "learning_rate": 9.989166070605447e-06,
+      "loss": 0.4966,
+      "step": 14
+    },
+    {
+      "epoch": 0.04491017964071856,
+      "grad_norm": 0.6413874138568662,
+      "learning_rate": 9.98756375606713e-06,
+      "loss": 0.463,
+      "step": 15
+    },
+    {
+      "epoch": 0.04790419161676647,
+      "grad_norm": 0.7098991626958999,
+      "learning_rate": 9.985851126551428e-06,
+      "loss": 0.4909,
+      "step": 16
+    },
+    {
+      "epoch": 0.05089820359281437,
+      "grad_norm": 0.5583816822067376,
+      "learning_rate": 9.9840282199383e-06,
+      "loss": 0.4019,
+      "step": 17
+    },
+    {
+      "epoch": 0.05389221556886228,
+      "grad_norm": 0.5749861156161733,
+      "learning_rate": 9.982095076546806e-06,
+      "loss": 0.4439,
+      "step": 18
+    },
+    {
+      "epoch": 0.05688622754491018,
+      "grad_norm": 0.6183025928753123,
+      "learning_rate": 9.980051739134235e-06,
+      "loss": 0.4589,
+      "step": 19
+    },
+    {
+      "epoch": 0.059880239520958084,
+      "grad_norm": 0.6169131746896298,
+      "learning_rate": 9.977898252895133e-06,
+      "loss": 0.4591,
+      "step": 20
+    },
+    {
+      "epoch": 0.06287425149700598,
+      "grad_norm": 0.5157803890102448,
+      "learning_rate": 9.975634665460333e-06,
+      "loss": 0.4343,
+      "step": 21
+    },
+    {
+      "epoch": 0.0658682634730539,
+      "grad_norm": 0.5120910534134643,
+      "learning_rate": 9.973261026895878e-06,
+      "loss": 0.391,
+      "step": 22
+    },
+    {
+      "epoch": 0.0688622754491018,
+      "grad_norm": 0.5373614499814718,
+      "learning_rate": 9.970777389701927e-06,
+      "loss": 0.482,
+      "step": 23
+    },
+    {
+      "epoch": 0.0718562874251497,
+      "grad_norm": 0.4841818267391823,
+      "learning_rate": 9.968183808811586e-06,
+      "loss": 0.3864,
+      "step": 24
+    },
+    {
+      "epoch": 0.0748502994011976,
+      "grad_norm": 0.613380603323421,
+      "learning_rate": 9.965480341589702e-06,
+      "loss": 0.449,
+      "step": 25
+    },
+    {
+      "epoch": 0.07784431137724551,
+      "grad_norm": 0.513581979497216,
+      "learning_rate": 9.962667047831585e-06,
+      "loss": 0.4486,
+      "step": 26
+    },
+    {
+      "epoch": 0.08083832335329341,
+      "grad_norm": 0.4858733602061576,
+      "learning_rate": 9.95974398976169e-06,
+      "loss": 0.3875,
+      "step": 27
+    },
+    {
+      "epoch": 0.08383233532934131,
+      "grad_norm": 0.5209865922036137,
+      "learning_rate": 9.95671123203224e-06,
+      "loss": 0.431,
+      "step": 28
+    },
+    {
+      "epoch": 0.08682634730538923,
+      "grad_norm": 0.5696963210781193,
+      "learning_rate": 9.953568841721796e-06,
+      "loss": 0.4447,
+      "step": 29
+    },
+    {
+      "epoch": 0.08982035928143713,
+      "grad_norm": 0.42441284052675404,
+      "learning_rate": 9.950316888333775e-06,
+      "loss": 0.3694,
+      "step": 30
+    },
+    {
+      "epoch": 0.09281437125748503,
+      "grad_norm": 0.43360808759085,
+      "learning_rate": 9.946955443794908e-06,
+      "loss": 0.4391,
+      "step": 31
+    },
+    {
+      "epoch": 0.09580838323353294,
+      "grad_norm": 0.5013748414124269,
+      "learning_rate": 9.943484582453653e-06,
+      "loss": 0.4091,
+      "step": 32
+    },
+    {
+      "epoch": 0.09880239520958084,
+      "grad_norm": 0.4419298351979592,
+      "learning_rate": 9.939904381078553e-06,
+      "loss": 0.3667,
+      "step": 33
+    },
+    {
+      "epoch": 0.10179640718562874,
+      "grad_norm": 0.39480170947609955,
+      "learning_rate": 9.93621491885653e-06,
+      "loss": 0.3401,
+      "step": 34
+    },
+    {
+      "epoch": 0.10479041916167664,
+      "grad_norm": 0.4933356814927667,
+      "learning_rate": 9.932416277391144e-06,
+      "loss": 0.3877,
+      "step": 35
+    },
+    {
+      "epoch": 0.10778443113772455,
+      "grad_norm": 0.4738646390646559,
+      "learning_rate": 9.928508540700775e-06,
+      "loss": 0.4069,
+      "step": 36
+    },
+    {
+      "epoch": 0.11077844311377245,
+      "grad_norm": 0.4823607402333716,
+      "learning_rate": 9.924491795216777e-06,
+      "loss": 0.4484,
+      "step": 37
+    },
+    {
+      "epoch": 0.11377245508982035,
+      "grad_norm": 0.48070642674174063,
+      "learning_rate": 9.920366129781564e-06,
+      "loss": 0.3979,
+      "step": 38
+    },
+    {
+      "epoch": 0.11676646706586827,
+      "grad_norm": 0.395354334469568,
+      "learning_rate": 9.916131635646635e-06,
+      "loss": 0.3592,
+      "step": 39
+    },
+    {
+      "epoch": 0.11976047904191617,
+      "grad_norm": 0.4579236551332394,
+      "learning_rate": 9.91178840647057e-06,
+      "loss": 0.4269,
+      "step": 40
+    },
+    {
+      "epoch": 0.12275449101796407,
+      "grad_norm": 0.4549107669701415,
+      "learning_rate": 9.907336538316946e-06,
+      "loss": 0.3791,
+      "step": 41
+    },
+    {
+      "epoch": 0.12574850299401197,
+      "grad_norm": 0.41412875854063724,
+      "learning_rate": 9.902776129652223e-06,
+      "loss": 0.3328,
+      "step": 42
+    },
+    {
+      "epoch": 0.12874251497005987,
+      "grad_norm": 0.45734106494684895,
+      "learning_rate": 9.898107281343557e-06,
+      "loss": 0.3426,
+      "step": 43
+    },
+    {
+      "epoch": 0.1317365269461078,
+      "grad_norm": 0.3975585171119778,
+      "learning_rate": 9.893330096656576e-06,
+      "loss": 0.3182,
+      "step": 44
+    },
+    {
+      "epoch": 0.1347305389221557,
+      "grad_norm": 0.4508251810720421,
+      "learning_rate": 9.888444681253087e-06,
+      "loss": 0.3961,
+      "step": 45
+    },
+    {
+      "epoch": 0.1377245508982036,
+      "grad_norm": 0.49018276906463715,
+      "learning_rate": 9.883451143188753e-06,
+      "loss": 0.3959,
+      "step": 46
+    },
+    {
+      "epoch": 0.1407185628742515,
+      "grad_norm": 0.43857108345958123,
+      "learning_rate": 9.878349592910694e-06,
+      "loss": 0.3223,
+      "step": 47
+    },
+    {
+      "epoch": 0.1437125748502994,
+      "grad_norm": 0.4914555045570781,
+      "learning_rate": 9.873140143255035e-06,
+      "loss": 0.4104,
+      "step": 48
+    },
+    {
+      "epoch": 0.1467065868263473,
+      "grad_norm": 0.4680416136318368,
+      "learning_rate": 9.867822909444435e-06,
+      "loss": 0.3992,
+      "step": 49
+    },
+    {
+      "epoch": 0.1497005988023952,
+      "grad_norm": 0.4464644127535241,
+      "learning_rate": 9.862398009085511e-06,
+      "loss": 0.3765,
+      "step": 50
+    },
+    {
+      "epoch": 0.15269461077844312,
+      "grad_norm": 0.44348514716100135,
+      "learning_rate": 9.856865562166256e-06,
+      "loss": 0.3554,
+      "step": 51
+    },
+    {
+      "epoch": 0.15568862275449102,
+      "grad_norm": 0.39090630445016866,
+      "learning_rate": 9.851225691053382e-06,
+      "loss": 0.3265,
+      "step": 52
+    },
+    {
+      "epoch": 0.15868263473053892,
+      "grad_norm": 0.4581711027478441,
+      "learning_rate": 9.8454785204896e-06,
+      "loss": 0.3944,
+      "step": 53
+    },
+    {
+      "epoch": 0.16167664670658682,
+      "grad_norm": 0.3718150929062592,
+      "learning_rate": 9.83962417759088e-06,
+      "loss": 0.3306,
+      "step": 54
+    },
+    {
+      "epoch": 0.16467065868263472,
+      "grad_norm": 0.46161900918940135,
+      "learning_rate": 9.833662791843628e-06,
+      "loss": 0.3393,
+      "step": 55
+    },
+    {
+      "epoch": 0.16766467065868262,
+      "grad_norm": 0.43249758867225013,
+      "learning_rate": 9.827594495101824e-06,
+      "loss": 0.3552,
+      "step": 56
+    },
+    {
+      "epoch": 0.17065868263473055,
+      "grad_norm": 0.42079987124693524,
+      "learning_rate": 9.821419421584108e-06,
+      "loss": 0.366,
+      "step": 57
+    },
+    {
+      "epoch": 0.17365269461077845,
+      "grad_norm": 0.42835319658565874,
+      "learning_rate": 9.815137707870806e-06,
+      "loss": 0.3897,
+      "step": 58
+    },
+    {
+      "epoch": 0.17664670658682635,
+      "grad_norm": 0.45378949632286464,
+      "learning_rate": 9.808749492900917e-06,
+      "loss": 0.3527,
+      "step": 59
+    },
+    {
+      "epoch": 0.17964071856287425,
+      "grad_norm": 0.41550341759356746,
+      "learning_rate": 9.802254917969033e-06,
+      "loss": 0.369,
+      "step": 60
+    },
+    {
+      "epoch": 0.18263473053892215,
+      "grad_norm": 0.47865365800877924,
+      "learning_rate": 9.795654126722218e-06,
+      "loss": 0.36,
+      "step": 61
+    },
+    {
+      "epoch": 0.18562874251497005,
+      "grad_norm": 0.41241747007098667,
+      "learning_rate": 9.788947265156828e-06,
+      "loss": 0.3441,
+      "step": 62
+    },
+    {
+      "epoch": 0.18862275449101795,
+      "grad_norm": 0.4252231101947087,
+      "learning_rate": 9.782134481615282e-06,
+      "loss": 0.3132,
+      "step": 63
+    },
+    {
+      "epoch": 0.19161676646706588,
+      "grad_norm": 0.42549981728492986,
+      "learning_rate": 9.775215926782788e-06,
+      "loss": 0.3781,
+      "step": 64
+    },
+    {
+      "epoch": 0.19461077844311378,
+      "grad_norm": 0.45200155726841834,
+      "learning_rate": 9.768191753683997e-06,
+      "loss": 0.3414,
+      "step": 65
+    },
+    {
+      "epoch": 0.19760479041916168,
+      "grad_norm": 0.43787571810405695,
+      "learning_rate": 9.761062117679632e-06,
+      "loss": 0.342,
+      "step": 66
+    },
+    {
+      "epoch": 0.20059880239520958,
+      "grad_norm": 0.4205549741056657,
+      "learning_rate": 9.75382717646304e-06,
+      "loss": 0.3222,
+      "step": 67
+    },
+    {
+      "epoch": 0.20359281437125748,
+      "grad_norm": 0.44297069302912406,
+      "learning_rate": 9.746487090056712e-06,
+      "loss": 0.4057,
+      "step": 68
+    },
+    {
+      "epoch": 0.20658682634730538,
+      "grad_norm": 0.36331306610217856,
+      "learning_rate": 9.739042020808746e-06,
+      "loss": 0.308,
+      "step": 69
+    },
+    {
+      "epoch": 0.20958083832335328,
+      "grad_norm": 0.4452426312638673,
+      "learning_rate": 9.73149213338924e-06,
+      "loss": 0.3831,
+      "step": 70
+    },
+    {
+      "epoch": 0.2125748502994012,
+      "grad_norm": 0.45916420452043477,
+      "learning_rate": 9.72383759478667e-06,
+      "loss": 0.3884,
+      "step": 71
+    },
+    {
+      "epoch": 0.2155688622754491,
+      "grad_norm": 0.42898892391628324,
+      "learning_rate": 9.71607857430419e-06,
+      "loss": 0.2995,
+      "step": 72
+    },
+    {
+      "epoch": 0.218562874251497,
+      "grad_norm": 0.46574867434425643,
+      "learning_rate": 9.708215243555875e-06,
+      "loss": 0.3736,
+      "step": 73
+    },
+    {
+      "epoch": 0.2215568862275449,
+      "grad_norm": 0.5109511095920206,
+      "learning_rate": 9.700247776462944e-06,
+      "loss": 0.3636,
+      "step": 74
+    },
+    {
+      "epoch": 0.2245508982035928,
+      "grad_norm": 0.4555316051552336,
+      "learning_rate": 9.6921763492499e-06,
+      "loss": 0.3463,
+      "step": 75
+    },
+    {
+      "epoch": 0.2275449101796407,
+      "grad_norm": 0.42251450705041066,
+      "learning_rate": 9.68400114044064e-06,
+      "loss": 0.2846,
+      "step": 76
+    },
+    {
+      "epoch": 0.23053892215568864,
+      "grad_norm": 0.4731483555029824,
+      "learning_rate": 9.6757223308545e-06,
+      "loss": 0.3563,
+      "step": 77
+    },
+    {
+      "epoch": 0.23353293413173654,
+      "grad_norm": 0.42993559776598106,
+      "learning_rate": 9.667340103602263e-06,
+      "loss": 0.4056,
+      "step": 78
+    },
+    {
+      "epoch": 0.23652694610778444,
+      "grad_norm": 0.4148151092265695,
+      "learning_rate": 9.658854644082099e-06,
+      "loss": 0.3226,
+      "step": 79
+    },
+    {
+      "epoch": 0.23952095808383234,
+      "grad_norm": 0.5414180623018148,
+      "learning_rate": 9.650266139975474e-06,
+      "loss": 0.3937,
+      "step": 80
+    },
+    {
+      "epoch": 0.24251497005988024,
+      "grad_norm": 0.4581306269118642,
+      "learning_rate": 9.641574781242999e-06,
+      "loss": 0.3319,
+      "step": 81
+    },
+    {
+      "epoch": 0.24550898203592814,
+      "grad_norm": 0.4722036512881197,
+      "learning_rate": 9.632780760120217e-06,
+      "loss": 0.3551,
+      "step": 82
+    },
+    {
+      "epoch": 0.24850299401197604,
+      "grad_norm": 0.442923177568666,
+      "learning_rate": 9.62388427111336e-06,
+      "loss": 0.32,
+      "step": 83
+    },
+    {
+      "epoch": 0.25149700598802394,
+      "grad_norm": 0.4105353863693248,
+      "learning_rate": 9.614885510995047e-06,
+      "loss": 0.2933,
+      "step": 84
+    },
+    {
+      "epoch": 0.25449101796407186,
+      "grad_norm": 0.4363298935780079,
+      "learning_rate": 9.605784678799934e-06,
+      "loss": 0.3384,
+      "step": 85
+    },
+    {
+      "epoch": 0.25748502994011974,
+      "grad_norm": 0.42752129079271195,
+      "learning_rate": 9.596581975820304e-06,
+      "loss": 0.3455,
+      "step": 86
+    },
+    {
+      "epoch": 0.26047904191616766,
+      "grad_norm": 0.452990009051003,
+      "learning_rate": 9.587277605601617e-06,
+      "loss": 0.3614,
+      "step": 87
+    },
+    {
+      "epoch": 0.2634730538922156,
+      "grad_norm": 0.5045675106659009,
+      "learning_rate": 9.577871773938013e-06,
+      "loss": 0.3866,
+      "step": 88
+    },
+    {
+      "epoch": 0.26646706586826346,
+      "grad_norm": 0.4756594194987398,
+      "learning_rate": 9.568364688867757e-06,
+      "loss": 0.4289,
+      "step": 89
+    },
+    {
+      "epoch": 0.2694610778443114,
+      "grad_norm": 0.42792935860102205,
+      "learning_rate": 9.558756560668637e-06,
+      "loss": 0.3278,
+      "step": 90
+    },
+    {
+      "epoch": 0.27245508982035926,
+      "grad_norm": 0.4626893456539531,
+      "learning_rate": 9.549047601853313e-06,
+      "loss": 0.341,
+      "step": 91
+    },
+    {
+      "epoch": 0.2754491017964072,
+      "grad_norm": 0.46215228342029285,
+      "learning_rate": 9.539238027164618e-06,
+      "loss": 0.3527,
+      "step": 92
+    },
+    {
+      "epoch": 0.27844311377245506,
+      "grad_norm": 0.4498517171769678,
+      "learning_rate": 9.52932805357081e-06,
+      "loss": 0.3468,
+      "step": 93
+    },
+    {
+      "epoch": 0.281437125748503,
+      "grad_norm": 0.4329877470458277,
+      "learning_rate": 9.519317900260769e-06,
+      "loss": 0.3662,
+      "step": 94
+    },
+    {
+      "epoch": 0.2844311377245509,
+      "grad_norm": 0.445828229536868,
+      "learning_rate": 9.509207788639148e-06,
+      "loss": 0.3598,
+      "step": 95
+    },
+    {
+      "epoch": 0.2874251497005988,
+      "grad_norm": 0.4259133675836244,
+      "learning_rate": 9.498997942321484e-06,
+      "loss": 0.3646,
+      "step": 96
+    },
+    {
+      "epoch": 0.2904191616766467,
+      "grad_norm": 0.442081394819526,
+      "learning_rate": 9.488688587129243e-06,
+      "loss": 0.3863,
+      "step": 97
+    },
+    {
+      "epoch": 0.2934131736526946,
+      "grad_norm": 0.47146521224998056,
+      "learning_rate": 9.47827995108483e-06,
+      "loss": 0.3226,
+      "step": 98
+    },
+    {
+      "epoch": 0.2964071856287425,
+      "grad_norm": 0.49360802389525715,
+      "learning_rate": 9.467772264406545e-06,
+      "loss": 0.3462,
+      "step": 99
+    },
+    {
+      "epoch": 0.2994011976047904,
+      "grad_norm": 0.3869276052014445,
+      "learning_rate": 9.457165759503492e-06,
+      "loss": 0.3151,
+      "step": 100
+    },
+    {
+      "epoch": 0.3023952095808383,
+      "grad_norm": 0.44144864606093637,
+      "learning_rate": 9.446460670970436e-06,
+      "loss": 0.32,
+      "step": 101
+    },
+    {
+      "epoch": 0.30538922155688625,
+      "grad_norm": 0.44635698043311317,
+      "learning_rate": 9.435657235582616e-06,
+      "loss": 0.2785,
+      "step": 102
+    },
+    {
+      "epoch": 0.3083832335329341,
+      "grad_norm": 0.47945916870819877,
+      "learning_rate": 9.424755692290507e-06,
+      "loss": 0.3444,
+      "step": 103
+    },
+    {
+      "epoch": 0.31137724550898205,
+      "grad_norm": 0.4889532067489411,
+      "learning_rate": 9.413756282214538e-06,
+      "loss": 0.3697,
+      "step": 104
+    },
+    {
+      "epoch": 0.3143712574850299,
+      "grad_norm": 0.4050374102899079,
+      "learning_rate": 9.402659248639749e-06,
+      "loss": 0.3391,
+      "step": 105
+    },
+    {
+      "epoch": 0.31736526946107785,
+      "grad_norm": 0.43134928270531775,
+      "learning_rate": 9.391464837010428e-06,
+      "loss": 0.3311,
+      "step": 106
+    },
+    {
+      "epoch": 0.3203592814371258,
+      "grad_norm": 0.4949908315827015,
+      "learning_rate": 9.380173294924661e-06,
+      "loss": 0.3784,
+      "step": 107
+    },
+    {
+      "epoch": 0.32335329341317365,
+      "grad_norm": 0.4245781514760333,
+      "learning_rate": 9.368784872128877e-06,
+      "loss": 0.3311,
+      "step": 108
+    },
+    {
+      "epoch": 0.3263473053892216,
+      "grad_norm": 0.4133547444227251,
+      "learning_rate": 9.357299820512305e-06,
+      "loss": 0.3221,
+      "step": 109
+    },
+    {
+      "epoch": 0.32934131736526945,
+      "grad_norm": 0.4444102279791174,
+      "learning_rate": 9.345718394101412e-06,
+      "loss": 0.3235,
+      "step": 110
+    },
+    {
+      "epoch": 0.3323353293413174,
+      "grad_norm": 0.45427034941731087,
+      "learning_rate": 9.334040849054288e-06,
+      "loss": 0.3579,
+      "step": 111
+    },
+    {
+      "epoch": 0.33532934131736525,
+      "grad_norm": 0.4487847778710975,
+      "learning_rate": 9.322267443654974e-06,
+      "loss": 0.3356,
+      "step": 112
+    },
+    {
+      "epoch": 0.3383233532934132,
+      "grad_norm": 0.4474444731493086,
+      "learning_rate": 9.310398438307747e-06,
+      "loss": 0.3811,
+      "step": 113
+    },
+    {
+      "epoch": 0.3413173652694611,
+      "grad_norm": 0.41875647883408995,
+      "learning_rate": 9.29843409553137e-06,
+      "loss": 0.3055,
+      "step": 114
+    },
+    {
+      "epoch": 0.344311377245509,
+      "grad_norm": 0.4094094286358269,
+      "learning_rate": 9.286374679953278e-06,
+      "loss": 0.322,
+      "step": 115
+    },
+    {
+      "epoch": 0.3473053892215569,
+      "grad_norm": 0.4176842888315912,
+      "learning_rate": 9.274220458303727e-06,
+      "loss": 0.2925,
+      "step": 116
+    },
+    {
+      "epoch": 0.3502994011976048,
+      "grad_norm": 0.4399995441523388,
+      "learning_rate": 9.261971699409893e-06,
+      "loss": 0.3439,
+      "step": 117
+    },
+    {
+      "epoch": 0.3532934131736527,
+      "grad_norm": 0.46862160672591463,
+      "learning_rate": 9.249628674189928e-06,
+      "loss": 0.3288,
+      "step": 118
+    },
+    {
+      "epoch": 0.3562874251497006,
+      "grad_norm": 0.47204157627676946,
+      "learning_rate": 9.237191655646972e-06,
+      "loss": 0.3632,
+      "step": 119
+    },
+    {
+      "epoch": 0.3592814371257485,
+      "grad_norm": 0.48318525696993125,
+      "learning_rate": 9.224660918863104e-06,
+      "loss": 0.3867,
+      "step": 120
+    },
+    {
+      "epoch": 0.36227544910179643,
+      "grad_norm": 0.4560878133746981,
+      "learning_rate": 9.212036740993265e-06,
+      "loss": 0.3338,
+      "step": 121
+    },
+    {
+      "epoch": 0.3652694610778443,
+      "grad_norm": 0.4587470203357078,
+      "learning_rate": 9.199319401259132e-06,
+      "loss": 0.3466,
+      "step": 122
+    },
+    {
+      "epoch": 0.36826347305389223,
+      "grad_norm": 0.4644666501367767,
+      "learning_rate": 9.186509180942928e-06,
+      "loss": 0.3227,
+      "step": 123
+    },
+    {
+      "epoch": 0.3712574850299401,
+      "grad_norm": 0.45285736689029926,
+      "learning_rate": 9.173606363381218e-06,
+      "loss": 0.379,
+      "step": 124
+    },
+    {
+      "epoch": 0.37425149700598803,
+      "grad_norm": 0.46989606724726096,
+      "learning_rate": 9.16061123395863e-06,
+      "loss": 0.3433,
+      "step": 125
+    },
+    {
+      "epoch": 0.3772455089820359,
+      "grad_norm": 0.4469910802103149,
+      "learning_rate": 9.147524080101543e-06,
+      "loss": 0.3352,
+      "step": 126
+    },
+    {
+      "epoch": 0.38023952095808383,
+      "grad_norm": 0.48805163979774363,
+      "learning_rate": 9.134345191271742e-06,
+      "loss": 0.3838,
+      "step": 127
+    },
+    {
+      "epoch": 0.38323353293413176,
+      "grad_norm": 0.4442468771716965,
+      "learning_rate": 9.121074858959997e-06,
+      "loss": 0.3633,
+      "step": 128
+    },
+    {
+      "epoch": 0.38622754491017963,
+      "grad_norm": 0.4047757062231612,
+      "learning_rate": 9.107713376679634e-06,
+      "loss": 0.3439,
+      "step": 129
+    },
+    {
+      "epoch": 0.38922155688622756,
+      "grad_norm": 0.48083208235155905,
+      "learning_rate": 9.094261039960028e-06,
+      "loss": 0.3475,
+      "step": 130
+    },
+    {
+      "epoch": 0.39221556886227543,
+      "grad_norm": 0.5289491347513543,
+      "learning_rate": 9.08071814634008e-06,
+      "loss": 0.399,
+      "step": 131
+    },
+    {
+      "epoch": 0.39520958083832336,
+      "grad_norm": 0.4682880868446404,
+      "learning_rate": 9.067084995361623e-06,
+      "loss": 0.3512,
+      "step": 132
+    },
+    {
+      "epoch": 0.39820359281437123,
+      "grad_norm": 0.47030940890948225,
+      "learning_rate": 9.053361888562807e-06,
+      "loss": 0.3815,
+      "step": 133
+    },
+    {
+      "epoch": 0.40119760479041916,
+      "grad_norm": 0.4266612160595512,
+      "learning_rate": 9.039549129471423e-06,
+      "loss": 0.3279,
+      "step": 134
+    },
+    {
+      "epoch": 0.4041916167664671,
+      "grad_norm": 0.3783107892151892,
+      "learning_rate": 9.025647023598196e-06,
+      "loss": 0.301,
+      "step": 135
+    },
+    {
+      "epoch": 0.40718562874251496,
+      "grad_norm": 0.4340730842843874,
+      "learning_rate": 9.011655878430018e-06,
+      "loss": 0.3146,
+      "step": 136
+    },
+    {
+      "epoch": 0.4101796407185629,
+      "grad_norm": 0.5008768268205231,
+      "learning_rate": 8.99757600342316e-06,
+      "loss": 0.3674,
+      "step": 137
+    },
+    {
+      "epoch": 0.41317365269461076,
+      "grad_norm": 0.5041973591950307,
+      "learning_rate": 8.983407709996415e-06,
+      "loss": 0.3554,
+      "step": 138
+    },
+    {
+      "epoch": 0.4161676646706587,
+      "grad_norm": 0.461219307672598,
+      "learning_rate": 8.969151311524215e-06,
+      "loss": 0.3254,
+      "step": 139
+    },
+    {
+      "epoch": 0.41916167664670656,
+      "grad_norm": 0.5245596751285111,
+      "learning_rate": 8.954807123329703e-06,
+      "loss": 0.3556,
+      "step": 140
+    },
+    {
+      "epoch": 0.4221556886227545,
+      "grad_norm": 0.389967119222637,
+      "learning_rate": 8.940375462677758e-06,
+      "loss": 0.2885,
+      "step": 141
+    },
+    {
+      "epoch": 0.4251497005988024,
+      "grad_norm": 0.46000170700343884,
+      "learning_rate": 8.92585664876797e-06,
+      "loss": 0.3224,
+      "step": 142
+    },
+    {
+      "epoch": 0.4281437125748503,
+      "grad_norm": 0.3901838814272129,
+      "learning_rate": 8.911251002727588e-06,
+      "loss": 0.3126,
+      "step": 143
+    },
+    {
+      "epoch": 0.4311377245508982,
+      "grad_norm": 0.41747104274916447,
+      "learning_rate": 8.896558847604414e-06,
+      "loss": 0.3053,
+      "step": 144
+    },
+    {
+      "epoch": 0.4341317365269461,
+      "grad_norm": 0.4469610609200074,
+      "learning_rate": 8.881780508359661e-06,
+      "loss": 0.4092,
+      "step": 145
+    },
+    {
+      "epoch": 0.437125748502994,
+      "grad_norm": 0.5576985825697979,
+      "learning_rate": 8.86691631186076e-06,
+      "loss": 0.3415,
+      "step": 146
+    },
+    {
+      "epoch": 0.44011976047904194,
+      "grad_norm": 0.37489263013835283,
+      "learning_rate": 8.851966586874138e-06,
+      "loss": 0.296,
+      "step": 147
+    },
+    {
+      "epoch": 0.4431137724550898,
+      "grad_norm": 0.4209073067612388,
+      "learning_rate": 8.836931664057935e-06,
+      "loss": 0.3125,
+      "step": 148
+    },
+    {
+      "epoch": 0.44610778443113774,
+      "grad_norm": 0.4908502136108264,
+      "learning_rate": 8.821811875954705e-06,
+      "loss": 0.3379,
+      "step": 149
+    },
+    {
+      "epoch": 0.4491017964071856,
+      "grad_norm": 0.4493204791244655,
+      "learning_rate": 8.806607556984045e-06,
+      "loss": 0.2961,
+      "step": 150
+    },
+    {
+      "epoch": 0.45209580838323354,
+      "grad_norm": 0.43468944452044395,
+      "learning_rate": 8.791319043435213e-06,
+      "loss": 0.3262,
+      "step": 151
+    },
+    {
+      "epoch": 0.4550898203592814,
+      "grad_norm": 0.40068706004376564,
+      "learning_rate": 8.775946673459682e-06,
+      "loss": 0.332,
+      "step": 152
+    },
+    {
+      "epoch": 0.45808383233532934,
+      "grad_norm": 0.4606909914110466,
+      "learning_rate": 8.76049078706366e-06,
+      "loss": 0.3077,
+      "step": 153
+    },
+    {
+      "epoch": 0.46107784431137727,
+      "grad_norm": 0.4263204989685283,
+      "learning_rate": 8.744951726100572e-06,
+      "loss": 0.2923,
+      "step": 154
+    },
+    {
+      "epoch": 0.46407185628742514,
+      "grad_norm": 0.49334469602827996,
+      "learning_rate": 8.729329834263503e-06,
+      "loss": 0.3167,
+      "step": 155
+    },
+    {
+      "epoch": 0.46706586826347307,
+      "grad_norm": 0.456124948055252,
+      "learning_rate": 8.713625457077585e-06,
+      "loss": 0.3471,
+      "step": 156
+    },
+    {
+      "epoch": 0.47005988023952094,
+      "grad_norm": 0.4797351378963682,
+      "learning_rate": 8.697838941892371e-06,
+      "loss": 0.3099,
+      "step": 157
+    },
+    {
+      "epoch": 0.47305389221556887,
+      "grad_norm": 0.4399550274160864,
+      "learning_rate": 8.681970637874131e-06,
+      "loss": 0.3567,
+      "step": 158
+    },
+    {
+      "epoch": 0.47604790419161674,
+      "grad_norm": 0.4540628971657261,
+      "learning_rate": 8.666020895998154e-06,
+      "loss": 0.3307,
+      "step": 159
+    },
+    {
+      "epoch": 0.47904191616766467,
+      "grad_norm": 0.4852212670098884,
+      "learning_rate": 8.64999006904096e-06,
+      "loss": 0.3477,
+      "step": 160
+    },
+    {
+      "epoch": 0.4820359281437126,
+      "grad_norm": 0.4759832601858189,
+      "learning_rate": 8.63387851157252e-06,
+      "loss": 0.3402,
+      "step": 161
+    },
+    {
+      "epoch": 0.48502994011976047,
+      "grad_norm": 0.4376051701589875,
+      "learning_rate": 8.617686579948396e-06,
+      "loss": 0.2942,
+      "step": 162
+    },
+    {
+      "epoch": 0.4880239520958084,
+      "grad_norm": 0.4827588897521435,
+      "learning_rate": 8.60141463230187e-06,
+      "loss": 0.3013,
+      "step": 163
+    },
+    {
+      "epoch": 0.49101796407185627,
+      "grad_norm": 0.50133948199433,
+      "learning_rate": 8.585063028536015e-06,
+      "loss": 0.3902,
+      "step": 164
+    },
+    {
+      "epoch": 0.4940119760479042,
+      "grad_norm": 0.432501330177427,
+      "learning_rate": 8.568632130315747e-06,
+      "loss": 0.3331,
+      "step": 165
+    },
+    {
+      "epoch": 0.49700598802395207,
+      "grad_norm": 0.4461230124416098,
+      "learning_rate": 8.552122301059807e-06,
+      "loss": 0.3264,
+      "step": 166
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.433268599961018,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 0.3121,
+      "step": 167
+    },
+    {
+      "epoch": 0.5029940119760479,
+      "grad_norm": 0.4360376926865427,
+      "learning_rate": 8.518867311836808e-06,
+      "loss": 0.3491,
+      "step": 168
+    },
+    {
+      "epoch": 0.5059880239520959,
+      "grad_norm": 0.48382769541626525,
+      "learning_rate": 8.502122887403882e-06,
+      "loss": 0.3624,
+      "step": 169
+    },
+    {
+      "epoch": 0.5089820359281437,
+      "grad_norm": 0.4827017650878896,
+      "learning_rate": 8.485301002987285e-06,
+      "loss": 0.2777,
+      "step": 170
+    },
+    {
+      "epoch": 0.5119760479041916,
+      "grad_norm": 0.4834475467045892,
+      "learning_rate": 8.468402030653598e-06,
+      "loss": 0.3767,
+      "step": 171
+    },
+    {
+      "epoch": 0.5149700598802395,
+      "grad_norm": 0.40832959634629307,
+      "learning_rate": 8.451426344174433e-06,
+      "loss": 0.309,
+      "step": 172
+    },
+    {
+      "epoch": 0.5179640718562875,
+      "grad_norm": 0.5028789702450074,
+      "learning_rate": 8.434374319018165e-06,
+      "loss": 0.3833,
+      "step": 173
+    },
+    {
+      "epoch": 0.5209580838323353,
+      "grad_norm": 0.428107703760953,
+      "learning_rate": 8.417246332341638e-06,
+      "loss": 0.3189,
+      "step": 174
+    },
+    {
+      "epoch": 0.5239520958083832,
+      "grad_norm": 0.43637985431285015,
+      "learning_rate": 8.4000427629818e-06,
+      "loss": 0.3264,
+      "step": 175
+    },
+    {
+      "epoch": 0.5269461077844312,
+      "grad_norm": 0.43684899444904945,
+      "learning_rate": 8.382763991447344e-06,
+      "loss": 0.357,
+      "step": 176
+    },
+    {
+      "epoch": 0.5299401197604791,
+      "grad_norm": 0.4678356778777381,
+      "learning_rate": 8.365410399910287e-06,
+      "loss": 0.335,
+      "step": 177
+    },
+    {
+      "epoch": 0.5329341317365269,
+      "grad_norm": 0.4853895362587734,
+      "learning_rate": 8.347982372197515e-06,
+      "loss": 0.3268,
+      "step": 178
+    },
+    {
+      "epoch": 0.5359281437125748,
+      "grad_norm": 0.43207052356131026,
+      "learning_rate": 8.33048029378229e-06,
+      "loss": 0.341,
+      "step": 179
+    },
+    {
+      "epoch": 0.5389221556886228,
+      "grad_norm": 0.398347323924377,
+      "learning_rate": 8.312904551775731e-06,
+      "loss": 0.2961,
+      "step": 180
+    },
+    {
+      "epoch": 0.5419161676646707,
+      "grad_norm": 0.4254629793234882,
+      "learning_rate": 8.295255534918249e-06,
+      "loss": 0.3299,
+      "step": 181
+    },
+    {
+      "epoch": 0.5449101796407185,
+      "grad_norm": 0.5328219429236702,
+      "learning_rate": 8.277533633570948e-06,
+      "loss": 0.3849,
+      "step": 182
+    },
+    {
+      "epoch": 0.5479041916167665,
+      "grad_norm": 0.4309732517050465,
+      "learning_rate": 8.25973923970699e-06,
+      "loss": 0.3543,
+      "step": 183
+    },
+    {
+      "epoch": 0.5508982035928144,
+      "grad_norm": 0.39450557734750547,
+      "learning_rate": 8.241872746902934e-06,
+      "loss": 0.394,
+      "step": 184
+    },
+    {
+      "epoch": 0.5538922155688623,
+      "grad_norm": 0.42038182562081156,
+      "learning_rate": 8.223934550330015e-06,
+      "loss": 0.3089,
+      "step": 185
+    },
+    {
+      "epoch": 0.5568862275449101,
+      "grad_norm": 0.4492225117294118,
+      "learning_rate": 8.20592504674542e-06,
+      "loss": 0.3608,
+      "step": 186
+    },
+    {
+      "epoch": 0.5598802395209581,
+      "grad_norm": 0.40302956852934263,
+      "learning_rate": 8.187844634483495e-06,
+      "loss": 0.3635,
+      "step": 187
+    },
+    {
+      "epoch": 0.562874251497006,
+      "grad_norm": 0.49411289366136724,
+      "learning_rate": 8.16969371344696e-06,
+      "loss": 0.4115,
+      "step": 188
+    },
+    {
+      "epoch": 0.5658682634730539,
+      "grad_norm": 0.38962051752449783,
+      "learning_rate": 8.151472685098037e-06,
+      "loss": 0.3318,
+      "step": 189
+    },
+    {
+      "epoch": 0.5688622754491018,
+      "grad_norm": 0.4196245005974281,
+      "learning_rate": 8.13318195244958e-06,
+      "loss": 0.3075,
+      "step": 190
+    },
+    {
+      "epoch": 0.5718562874251497,
+      "grad_norm": 0.43963849415324896,
+      "learning_rate": 8.114821920056177e-06,
+      "loss": 0.3366,
+      "step": 191
+    },
+    {
+      "epoch": 0.5748502994011976,
+      "grad_norm": 0.41549872294637885,
+      "learning_rate": 8.096392994005177e-06,
+      "loss": 0.3549,
+      "step": 192
+    },
+    {
+      "epoch": 0.5778443113772455,
+      "grad_norm": 0.4050502597513978,
+      "learning_rate": 8.077895581907719e-06,
+      "loss": 0.3305,
+      "step": 193
+    },
+    {
+      "epoch": 0.5808383233532934,
+      "grad_norm": 0.37822382244523783,
+      "learning_rate": 8.059330092889724e-06,
+      "loss": 0.281,
+      "step": 194
+    },
+    {
+      "epoch": 0.5838323353293413,
+      "grad_norm": 0.43683053908450975,
+      "learning_rate": 8.040696937582833e-06,
+      "loss": 0.3275,
+      "step": 195
+    },
+    {
+      "epoch": 0.5868263473053892,
+      "grad_norm": 0.44330247907511044,
+      "learning_rate": 8.021996528115335e-06,
+      "loss": 0.2985,
+      "step": 196
+    },
+    {
+      "epoch": 0.5898203592814372,
+      "grad_norm": 0.4527540758903123,
+      "learning_rate": 8.003229278103044e-06,
+      "loss": 0.3165,
+      "step": 197
+    },
+    {
+      "epoch": 0.592814371257485,
+      "grad_norm": 0.43916164453369716,
+      "learning_rate": 7.984395602640153e-06,
+      "loss": 0.3134,
+      "step": 198
+    },
+    {
+      "epoch": 0.5958083832335329,
+      "grad_norm": 0.49082807954078456,
+      "learning_rate": 7.96549591829006e-06,
+      "loss": 0.3314,
+      "step": 199
+    },
+    {
+      "epoch": 0.5988023952095808,
+      "grad_norm": 0.3727943752321762,
+      "learning_rate": 7.946530643076138e-06,
+      "loss": 0.2796,
+      "step": 200
+    },
+    {
+      "epoch": 0.5988023952095808,
+      "eval_loss": 0.3253346383571625,
+      "eval_runtime": 2.3286,
+      "eval_samples_per_second": 23.19,
+      "eval_steps_per_second": 6.012,
+      "step": 200
+    },
+    {
+      "epoch": 0.6017964071856288,
+      "grad_norm": 0.42666034708665596,
+      "learning_rate": 7.927500196472506e-06,
+      "loss": 0.3228,
+      "step": 201
+    },
+    {
+      "epoch": 0.6047904191616766,
+      "grad_norm": 0.4409784999155233,
+      "learning_rate": 7.908404999394747e-06,
+      "loss": 0.2799,
+      "step": 202
+    },
+    {
+      "epoch": 0.6077844311377245,
+      "grad_norm": 0.4266124452926282,
+      "learning_rate": 7.889245474190588e-06,
+      "loss": 0.3209,
+      "step": 203
+    },
+    {
+      "epoch": 0.6107784431137725,
+      "grad_norm": 0.5147186678748065,
+      "learning_rate": 7.870022044630569e-06,
+      "loss": 0.3559,
+      "step": 204
+    },
+    {
+      "epoch": 0.6137724550898204,
+      "grad_norm": 0.4165713982615347,
+      "learning_rate": 7.85073513589867e-06,
+      "loss": 0.3271,
+      "step": 205
+    },
+    {
+      "epoch": 0.6167664670658682,
+      "grad_norm": 0.5443853074634595,
+      "learning_rate": 7.831385174582901e-06,
+      "loss": 0.3311,
+      "step": 206
+    },
+    {
+      "epoch": 0.6197604790419161,
+      "grad_norm": 0.43530225836084246,
+      "learning_rate": 7.81197258866587e-06,
+      "loss": 0.3375,
+      "step": 207
+    },
+    {
+      "epoch": 0.6227544910179641,
+      "grad_norm": 0.42758182069776285,
+      "learning_rate": 7.792497807515317e-06,
+      "loss": 0.2946,
+      "step": 208
+    },
+    {
+      "epoch": 0.625748502994012,
+      "grad_norm": 0.44650284684212554,
+      "learning_rate": 7.772961261874615e-06,
+      "loss": 0.2857,
+      "step": 209
+    },
+    {
+      "epoch": 0.6287425149700598,
+      "grad_norm": 0.4270664729799722,
+      "learning_rate": 7.75336338385325e-06,
+      "loss": 0.3117,
+      "step": 210
+    },
+    {
+      "epoch": 0.6317365269461078,
+      "grad_norm": 0.45512231386199486,
+      "learning_rate": 7.733704606917248e-06,
+      "loss": 0.3441,
+      "step": 211
+    },
+    {
+      "epoch": 0.6347305389221557,
+      "grad_norm": 0.4542245407355785,
+      "learning_rate": 7.713985365879607e-06,
+      "loss": 0.3002,
+      "step": 212
+    },
+    {
+      "epoch": 0.6377245508982036,
+      "grad_norm": 0.4286759680111227,
+      "learning_rate": 7.694206096890667e-06,
+      "loss": 0.3021,
+      "step": 213
+    },
+    {
+      "epoch": 0.6407185628742516,
+      "grad_norm": 0.4054565768342119,
+      "learning_rate": 7.674367237428467e-06,
+      "loss": 0.2909,
+      "step": 214
+    },
+    {
+      "epoch": 0.6437125748502994,
+      "grad_norm": 0.3992967767133553,
+      "learning_rate": 7.654469226289068e-06,
+      "loss": 0.2725,
+      "step": 215
+    },
+    {
+      "epoch": 0.6467065868263473,
+      "grad_norm": 0.44575321770091864,
+      "learning_rate": 7.63451250357685e-06,
+      "loss": 0.3385,
+      "step": 216
+    },
+    {
+      "epoch": 0.6497005988023952,
+      "grad_norm": 0.43952231736992686,
+      "learning_rate": 7.614497510694774e-06,
+      "loss": 0.3405,
+      "step": 217
+    },
+    {
+      "epoch": 0.6526946107784432,
+      "grad_norm": 0.4709549657173294,
+      "learning_rate": 7.5944246903346204e-06,
+      "loss": 0.3022,
+      "step": 218
+    },
+    {
+      "epoch": 0.655688622754491,
+      "grad_norm": 0.40837282664768926,
+      "learning_rate": 7.574294486467204e-06,
+      "loss": 0.3139,
+      "step": 219
+    },
+    {
+      "epoch": 0.6586826347305389,
+      "grad_norm": 0.451012793296954,
+      "learning_rate": 7.55410734433254e-06,
+      "loss": 0.3475,
+      "step": 220
+    },
+    {
+      "epoch": 0.6616766467065869,
+      "grad_norm": 0.4890992121477276,
+      "learning_rate": 7.533863710430011e-06,
+      "loss": 0.3607,
+      "step": 221
+    },
+    {
+      "epoch": 0.6646706586826348,
+      "grad_norm": 0.41890430377213406,
+      "learning_rate": 7.513564032508484e-06,
+      "loss": 0.2963,
+      "step": 222
+    },
+    {
+      "epoch": 0.6676646706586826,
+      "grad_norm": 0.4700620020156656,
+      "learning_rate": 7.493208759556406e-06,
+      "loss": 0.3529,
+      "step": 223
+    },
+    {
+      "epoch": 0.6706586826347305,
+      "grad_norm": 0.5217871737505592,
+      "learning_rate": 7.472798341791877e-06,
+      "loss": 0.3728,
+      "step": 224
+    },
+    {
+      "epoch": 0.6736526946107785,
+      "grad_norm": 0.40835853071005473,
+      "learning_rate": 7.452333230652688e-06,
+      "loss": 0.3071,
+      "step": 225
+    },
+    {
+      "epoch": 0.6766467065868264,
+      "grad_norm": 0.3851499538311077,
+      "learning_rate": 7.431813878786343e-06,
+      "loss": 0.3037,
+      "step": 226
+    },
+    {
+      "epoch": 0.6796407185628742,
+      "grad_norm": 0.4565111849050823,
+      "learning_rate": 7.4112407400400395e-06,
+      "loss": 0.3444,
+      "step": 227
+    },
+    {
+      "epoch": 0.6826347305389222,
+      "grad_norm": 0.4393128646002543,
+      "learning_rate": 7.390614269450633e-06,
+      "loss": 0.323,
+      "step": 228
+    },
+    {
+      "epoch": 0.6856287425149701,
+      "grad_norm": 0.3953778614838876,
+      "learning_rate": 7.369934923234577e-06,
+      "loss": 0.301,
+      "step": 229
+    },
+    {
+      "epoch": 0.688622754491018,
+      "grad_norm": 0.49063863694418025,
+      "learning_rate": 7.349203158777826e-06,
+      "loss": 0.3594,
+      "step": 230
+    },
+    {
+      "epoch": 0.6916167664670658,
+      "grad_norm": 0.45489295228927934,
+      "learning_rate": 7.32841943462572e-06,
+      "loss": 0.329,
+      "step": 231
+    },
+    {
+      "epoch": 0.6946107784431138,
+      "grad_norm": 0.45527591702392517,
+      "learning_rate": 7.3075842104728445e-06,
+      "loss": 0.3388,
+      "step": 232
+    },
+    {
+      "epoch": 0.6976047904191617,
+      "grad_norm": 0.4533179025176441,
+      "learning_rate": 7.286697947152868e-06,
+      "loss": 0.3518,
+      "step": 233
+    },
+    {
+      "epoch": 0.7005988023952096,
+      "grad_norm": 0.4567400190803045,
+      "learning_rate": 7.265761106628338e-06,
+      "loss": 0.3146,
+      "step": 234
+    },
+    {
+      "epoch": 0.7035928143712575,
+      "grad_norm": 0.4096145410878116,
+      "learning_rate": 7.244774151980466e-06,
+      "loss": 0.3074,
+      "step": 235
+    },
+    {
+      "epoch": 0.7065868263473054,
+      "grad_norm": 0.5401969822391114,
+      "learning_rate": 7.223737547398898e-06,
+      "loss": 0.3117,
+      "step": 236
+    },
+    {
+      "epoch": 0.7095808383233533,
+      "grad_norm": 0.4456799566555001,
+      "learning_rate": 7.20265175817143e-06,
+      "loss": 0.3259,
+      "step": 237
+    },
+    {
+      "epoch": 0.7125748502994012,
+      "grad_norm": 0.49596678781262976,
+      "learning_rate": 7.181517250673729e-06,
+      "loss": 0.3567,
+      "step": 238
+    },
+    {
+      "epoch": 0.7155688622754491,
+      "grad_norm": 0.4614502461314865,
+      "learning_rate": 7.1603344923590065e-06,
+      "loss": 0.3134,
+      "step": 239
+    },
+    {
+      "epoch": 0.718562874251497,
+      "grad_norm": 0.424217638868852,
+      "learning_rate": 7.139103951747694e-06,
+      "loss": 0.3443,
+      "step": 240
+    },
+    {
+      "epoch": 0.7215568862275449,
+      "grad_norm": 0.4169590555256484,
+      "learning_rate": 7.1178260984170675e-06,
+      "loss": 0.2919,
+      "step": 241
+    },
+    {
+      "epoch": 0.7245508982035929,
+      "grad_norm": 0.4151427176467248,
+      "learning_rate": 7.0965014029908654e-06,
+      "loss": 0.2907,
+      "step": 242
+    },
+    {
+      "epoch": 0.7275449101796407,
+      "grad_norm": 0.38016547923897637,
+      "learning_rate": 7.075130337128883e-06,
+      "loss": 0.3172,
+      "step": 243
+    },
+    {
+      "epoch": 0.7305389221556886,
+      "grad_norm": 0.4560630496508005,
+      "learning_rate": 7.053713373516538e-06,
+      "loss": 0.3368,
+      "step": 244
+    },
+    {
+      "epoch": 0.7335329341317365,
+      "grad_norm": 0.41307088658462027,
+      "learning_rate": 7.03225098585441e-06,
+      "loss": 0.3437,
+      "step": 245
+    },
+    {
+      "epoch": 0.7365269461077845,
+      "grad_norm": 0.3993536350264866,
+      "learning_rate": 7.0107436488477694e-06,
+      "loss": 0.3269,
+      "step": 246
+    },
+    {
+      "epoch": 0.7395209580838323,
+      "grad_norm": 0.4390713120175351,
+      "learning_rate": 6.989191838196083e-06,
+      "loss": 0.3494,
+      "step": 247
+    },
+    {
+      "epoch": 0.7425149700598802,
+      "grad_norm": 0.42040069424471355,
+      "learning_rate": 6.9675960305824785e-06,
+      "loss": 0.3031,
+      "step": 248
+    },
+    {
+      "epoch": 0.7455089820359282,
+      "grad_norm": 0.5034439306976819,
+      "learning_rate": 6.945956703663212e-06,
+      "loss": 0.3373,
+      "step": 249
+    },
+    {
+      "epoch": 0.7485029940119761,
+      "grad_norm": 0.39661140948569246,
+      "learning_rate": 6.9242743360570985e-06,
+      "loss": 0.3082,
+      "step": 250
+    },
+    {
+      "epoch": 0.7514970059880239,
+      "grad_norm": 0.49981001871065467,
+      "learning_rate": 6.9025494073349284e-06,
+      "loss": 0.36,
+      "step": 251
+    },
+    {
+      "epoch": 0.7544910179640718,
+      "grad_norm": 0.41173662088074203,
+      "learning_rate": 6.880782398008862e-06,
+      "loss": 0.3243,
+      "step": 252
+    },
+    {
+      "epoch": 0.7574850299401198,
+      "grad_norm": 0.49925548773448797,
+      "learning_rate": 6.858973789521792e-06,
+      "loss": 0.2895,
+      "step": 253
+    },
+    {
+      "epoch": 0.7604790419161677,
+      "grad_norm": 0.4606637821324601,
+      "learning_rate": 6.837124064236709e-06,
+      "loss": 0.3558,
+      "step": 254
+    },
+    {
+      "epoch": 0.7634730538922155,
+      "grad_norm": 0.4158572209872501,
+      "learning_rate": 6.815233705426019e-06,
+      "loss": 0.3212,
+      "step": 255
+    },
+    {
+      "epoch": 0.7664670658682635,
+      "grad_norm": 0.46456116062331415,
+      "learning_rate": 6.7933031972608644e-06,
+      "loss": 0.3507,
+      "step": 256
+    },
+    {
+      "epoch": 0.7694610778443114,
+      "grad_norm": 0.42346412793061133,
+      "learning_rate": 6.771333024800411e-06,
+      "loss": 0.3374,
+      "step": 257
+    },
+    {
+      "epoch": 0.7724550898203593,
+      "grad_norm": 0.5308253026055302,
+      "learning_rate": 6.74932367398112e-06,
+      "loss": 0.3227,
+      "step": 258
+    },
+    {
+      "epoch": 0.7754491017964071,
+      "grad_norm": 0.4247577092823958,
+      "learning_rate": 6.727275631605996e-06,
+      "loss": 0.2901,
+      "step": 259
+    },
+    {
+      "epoch": 0.7784431137724551,
+      "grad_norm": 0.4514258503959119,
+      "learning_rate": 6.70518938533383e-06,
+      "loss": 0.3192,
+      "step": 260
+    },
+    {
+      "epoch": 0.781437125748503,
+      "grad_norm": 0.45592673448760346,
+      "learning_rate": 6.683065423668403e-06,
+      "loss": 0.3411,
+      "step": 261
+    },
+    {
+      "epoch": 0.7844311377245509,
+      "grad_norm": 0.45582223191458204,
+      "learning_rate": 6.660904235947687e-06,
+      "loss": 0.3206,
+      "step": 262
+    },
+    {
+      "epoch": 0.7874251497005988,
+      "grad_norm": 0.4464790548941145,
+      "learning_rate": 6.638706312333018e-06,
+      "loss": 0.3447,
+      "step": 263
+    },
+    {
+      "epoch": 0.7904191616766467,
+      "grad_norm": 0.4613341680799862,
+      "learning_rate": 6.61647214379826e-06,
+      "loss": 0.3731,
+      "step": 264
+    },
+    {
+      "epoch": 0.7934131736526946,
+      "grad_norm": 0.525401617770996,
+      "learning_rate": 6.594202222118941e-06,
+      "loss": 0.3395,
+      "step": 265
+    },
+    {
+      "epoch": 0.7964071856287425,
+      "grad_norm": 0.4768157641567023,
+      "learning_rate": 6.571897039861377e-06,
+      "loss": 0.3194,
+      "step": 266
+    },
+    {
+      "epoch": 0.7994011976047904,
+      "grad_norm": 0.42615524105225905,
+      "learning_rate": 6.549557090371775e-06,
+      "loss": 0.3305,
+      "step": 267
+    },
+    {
+      "epoch": 0.8023952095808383,
+      "grad_norm": 0.40204984708895913,
+      "learning_rate": 6.527182867765333e-06,
+      "loss": 0.2435,
+      "step": 268
+    },
+    {
+      "epoch": 0.8053892215568862,
+      "grad_norm": 0.45721155985991097,
+      "learning_rate": 6.504774866915291e-06,
+      "loss": 0.3114,
+      "step": 269
+    },
+    {
+      "epoch": 0.8083832335329342,
+      "grad_norm": 0.5019917137001293,
+      "learning_rate": 6.482333583442002e-06,
+      "loss": 0.2947,
+      "step": 270
+    },
+    {
+      "epoch": 0.811377245508982,
+      "grad_norm": 0.40166857362866804,
+      "learning_rate": 6.459859513701967e-06,
+      "loss": 0.3227,
+      "step": 271
+    },
+    {
+      "epoch": 0.8143712574850299,
+      "grad_norm": 0.5090510186869257,
+      "learning_rate": 6.437353154776848e-06,
+      "loss": 0.3278,
+      "step": 272
+    },
+    {
+      "epoch": 0.8173652694610778,
+      "grad_norm": 0.489599920238901,
+      "learning_rate": 6.414815004462483e-06,
+      "loss": 0.3527,
+      "step": 273
+    },
+    {
+      "epoch": 0.8203592814371258,
+      "grad_norm": 0.43141165848277235,
+      "learning_rate": 6.3922455612578715e-06,
+      "loss": 0.354,
+      "step": 274
+    },
+    {
+      "epoch": 0.8233532934131736,
+      "grad_norm": 0.43826955516164745,
+      "learning_rate": 6.369645324354149e-06,
+      "loss": 0.2743,
+      "step": 275
+    },
+    {
+      "epoch": 0.8263473053892215,
+      "grad_norm": 0.41461420614840017,
+      "learning_rate": 6.3470147936235485e-06,
+      "loss": 0.3283,
+      "step": 276
+    },
+    {
+      "epoch": 0.8293413173652695,
+      "grad_norm": 0.3576229329688004,
+      "learning_rate": 6.3243544696083355e-06,
+      "loss": 0.2885,
+      "step": 277
+    },
+    {
+      "epoch": 0.8323353293413174,
+      "grad_norm": 0.407950598883862,
+      "learning_rate": 6.301664853509755e-06,
+      "loss": 0.2793,
+      "step": 278
+    },
+    {
+      "epoch": 0.8353293413173652,
+      "grad_norm": 0.38807232592625707,
+      "learning_rate": 6.278946447176924e-06,
+      "loss": 0.3157,
+      "step": 279
+    },
+    {
+      "epoch": 0.8383233532934131,
+      "grad_norm": 0.39870061301820325,
+      "learning_rate": 6.256199753095745e-06,
+      "loss": 0.293,
+      "step": 280
+    },
+    {
+      "epoch": 0.8413173652694611,
+      "grad_norm": 0.41337239676594406,
+      "learning_rate": 6.233425274377793e-06,
+      "loss": 0.3095,
+      "step": 281
+    },
+    {
+      "epoch": 0.844311377245509,
+      "grad_norm": 0.5421319795588267,
+      "learning_rate": 6.21062351474918e-06,
+      "loss": 0.3626,
+      "step": 282
+    },
+    {
+      "epoch": 0.8473053892215568,
+      "grad_norm": 0.43090230373986216,
+      "learning_rate": 6.18779497853942e-06,
+      "loss": 0.3391,
+      "step": 283
+    },
+    {
+      "epoch": 0.8502994011976048,
+      "grad_norm": 0.4311618058945252,
+      "learning_rate": 6.164940170670266e-06,
+      "loss": 0.3447,
+      "step": 284
+    },
+    {
+      "epoch": 0.8532934131736527,
+      "grad_norm": 0.3630507384729582,
+      "learning_rate": 6.142059596644557e-06,
+      "loss": 0.2684,
+      "step": 285
+    },
+    {
+      "epoch": 0.8562874251497006,
+      "grad_norm": 0.44399389237149617,
+      "learning_rate": 6.11915376253502e-06,
+      "loss": 0.3004,
+      "step": 286
+    },
+    {
+      "epoch": 0.8592814371257484,
+      "grad_norm": 0.4446424601717037,
+      "learning_rate": 6.096223174973091e-06,
+      "loss": 0.3263,
+      "step": 287
+    },
+    {
+      "epoch": 0.8622754491017964,
+      "grad_norm": 0.386025094285913,
+      "learning_rate": 6.073268341137694e-06,
+      "loss": 0.2708,
+      "step": 288
+    },
+    {
+      "epoch": 0.8652694610778443,
+      "grad_norm": 0.45032311964136956,
+      "learning_rate": 6.050289768744042e-06,
+      "loss": 0.2663,
+      "step": 289
+    },
+    {
+      "epoch": 0.8682634730538922,
+      "grad_norm": 0.4748592521975155,
+      "learning_rate": 6.0272879660323936e-06,
+      "loss": 0.3517,
+      "step": 290
+    },
+    {
+      "epoch": 0.8712574850299402,
+      "grad_norm": 0.4353585226442709,
+      "learning_rate": 6.004263441756815e-06,
+      "loss": 0.3242,
+      "step": 291
+    },
+    {
+      "epoch": 0.874251497005988,
+      "grad_norm": 0.459994500506146,
+      "learning_rate": 5.98121670517393e-06,
+      "loss": 0.3063,
+      "step": 292
+    },
+    {
+      "epoch": 0.8772455089820359,
+      "grad_norm": 0.3719416324918383,
+      "learning_rate": 5.958148266031654e-06,
+      "loss": 0.2737,
+      "step": 293
+    },
+    {
+      "epoch": 0.8802395209580839,
+      "grad_norm": 0.4133018023008959,
+      "learning_rate": 5.935058634557917e-06,
+      "loss": 0.2916,
+      "step": 294
+    },
+    {
+      "epoch": 0.8832335329341318,
+      "grad_norm": 0.38458573094833726,
+      "learning_rate": 5.911948321449384e-06,
+      "loss": 0.324,
+      "step": 295
+    },
+    {
+      "epoch": 0.8862275449101796,
+      "grad_norm": 0.4661831583100884,
+      "learning_rate": 5.8888178378601565e-06,
+      "loss": 0.3319,
+      "step": 296
+    },
+    {
+      "epoch": 0.8892215568862275,
+      "grad_norm": 0.4590664280729926,
+      "learning_rate": 5.865667695390468e-06,
+      "loss": 0.3168,
+      "step": 297
+    },
+    {
+      "epoch": 0.8922155688622755,
+      "grad_norm": 0.4171407892831031,
+      "learning_rate": 5.842498406075363e-06,
+      "loss": 0.2992,
+      "step": 298
+    },
+    {
+      "epoch": 0.8952095808383234,
+      "grad_norm": 0.44453590562402706,
+      "learning_rate": 5.819310482373381e-06,
+      "loss": 0.319,
+      "step": 299
+    },
+    {
+      "epoch": 0.8982035928143712,
+      "grad_norm": 0.4309415928792554,
+      "learning_rate": 5.796104437155213e-06,
+      "loss": 0.3502,
+      "step": 300
+    },
+    {
+      "epoch": 0.9011976047904192,
+      "grad_norm": 0.4529128201598535,
+      "learning_rate": 5.772880783692363e-06,
+      "loss": 0.2933,
+      "step": 301
+    },
+    {
+      "epoch": 0.9041916167664671,
+      "grad_norm": 0.4838900364874081,
+      "learning_rate": 5.749640035645798e-06,
+      "loss": 0.2846,
+      "step": 302
+    },
+    {
+      "epoch": 0.907185628742515,
+      "grad_norm": 0.46961105639562856,
+      "learning_rate": 5.726382707054578e-06,
+      "loss": 0.3179,
+      "step": 303
+    },
+    {
+      "epoch": 0.9101796407185628,
+      "grad_norm": 0.4374574596846909,
+      "learning_rate": 5.703109312324493e-06,
+      "loss": 0.3061,
+      "step": 304
+    },
+    {
+      "epoch": 0.9131736526946108,
+      "grad_norm": 0.4790846572931936,
+      "learning_rate": 5.679820366216684e-06,
+      "loss": 0.3737,
+      "step": 305
+    },
+    {
+      "epoch": 0.9161676646706587,
+      "grad_norm": 0.39669367332017963,
+      "learning_rate": 5.656516383836263e-06,
+      "loss": 0.2722,
+      "step": 306
+    },
+    {
+      "epoch": 0.9191616766467066,
+      "grad_norm": 0.4433288165777792,
+      "learning_rate": 5.6331978806209044e-06,
+      "loss": 0.3246,
+      "step": 307
+    },
+    {
+      "epoch": 0.9221556886227545,
+      "grad_norm": 0.44706188719051204,
+      "learning_rate": 5.609865372329461e-06,
+      "loss": 0.2938,
+      "step": 308
+    },
+    {
+      "epoch": 0.9251497005988024,
+      "grad_norm": 0.4047628681820038,
+      "learning_rate": 5.586519375030549e-06,
+      "loss": 0.2646,
+      "step": 309
+    },
+    {
+      "epoch": 0.9281437125748503,
+      "grad_norm": 0.40013007673621875,
+      "learning_rate": 5.5631604050911354e-06,
+      "loss": 0.2947,
+      "step": 310
+    },
+    {
+      "epoch": 0.9311377245508982,
+      "grad_norm": 0.49696203365829383,
+      "learning_rate": 5.539788979165115e-06,
+      "loss": 0.2795,
+      "step": 311
+    },
+    {
+      "epoch": 0.9341317365269461,
+      "grad_norm": 0.3854919443524009,
+      "learning_rate": 5.516405614181883e-06,
+      "loss": 0.2606,
+      "step": 312
+    },
+    {
+      "epoch": 0.937125748502994,
+      "grad_norm": 0.3951202996569865,
+      "learning_rate": 5.4930108273349034e-06,
+      "loss": 0.2988,
+      "step": 313
+    },
+    {
+      "epoch": 0.9401197604790419,
+      "grad_norm": 0.43356465525658267,
+      "learning_rate": 5.4696051360702725e-06,
+      "loss": 0.2809,
+      "step": 314
+    },
+    {
+      "epoch": 0.9431137724550899,
+      "grad_norm": 0.48252252235411114,
+      "learning_rate": 5.446189058075265e-06,
+      "loss": 0.3295,
+      "step": 315
+    },
+    {
+      "epoch": 0.9461077844311377,
+      "grad_norm": 0.4457754507758648,
+      "learning_rate": 5.4227631112668955e-06,
+      "loss": 0.329,
+      "step": 316
+    },
+    {
+      "epoch": 0.9491017964071856,
+      "grad_norm": 0.4563222170680985,
+      "learning_rate": 5.39932781378045e-06,
+      "loss": 0.3411,
+      "step": 317
+    },
+    {
+      "epoch": 0.9520958083832335,
+      "grad_norm": 0.3817629536413549,
+      "learning_rate": 5.375883683958041e-06,
+      "loss": 0.3008,
+      "step": 318
+    },
+    {
+      "epoch": 0.9550898203592815,
+      "grad_norm": 0.4428161166333622,
+      "learning_rate": 5.3524312403371255e-06,
+      "loss": 0.3121,
+      "step": 319
+    },
+    {
+      "epoch": 0.9580838323353293,
+      "grad_norm": 0.4800225224883845,
+      "learning_rate": 5.328971001639054e-06,
+      "loss": 0.2793,
+      "step": 320
+    },
+    {
+      "epoch": 0.9610778443113772,
+      "grad_norm": 0.42308169816478763,
+      "learning_rate": 5.3055034867575825e-06,
+      "loss": 0.2821,
+      "step": 321
+    },
+    {
+      "epoch": 0.9640718562874252,
+      "grad_norm": 0.5069133125830388,
+      "learning_rate": 5.282029214747404e-06,
+      "loss": 0.3335,
+      "step": 322
+    },
+    {
+      "epoch": 0.9670658682634731,
+      "grad_norm": 0.4294282858011169,
+      "learning_rate": 5.258548704812667e-06,
+      "loss": 0.3238,
+      "step": 323
+    },
+    {
+      "epoch": 0.9700598802395209,
+      "grad_norm": 0.4304895993240004,
+      "learning_rate": 5.235062476295488e-06,
+      "loss": 0.3037,
+      "step": 324
+    },
+    {
+      "epoch": 0.9730538922155688,
+      "grad_norm": 0.44665184518929646,
+      "learning_rate": 5.211571048664469e-06,
+      "loss": 0.2681,
+      "step": 325
+    },
+    {
+      "epoch": 0.9760479041916168,
+      "grad_norm": 0.4194880264910065,
+      "learning_rate": 5.188074941503203e-06,
+      "loss": 0.3262,
+      "step": 326
+    },
+    {
+      "epoch": 0.9790419161676647,
+      "grad_norm": 0.43775778284917766,
+      "learning_rate": 5.164574674498788e-06,
+      "loss": 0.3192,
+      "step": 327
+    },
+    {
+      "epoch": 0.9820359281437125,
+      "grad_norm": 0.48137735332553216,
+      "learning_rate": 5.141070767430331e-06,
+      "loss": 0.3068,
+      "step": 328
+    },
+    {
+      "epoch": 0.9850299401197605,
+      "grad_norm": 0.4300785083365142,
+      "learning_rate": 5.117563740157444e-06,
+      "loss": 0.2958,
+      "step": 329
+    },
+    {
+      "epoch": 0.9880239520958084,
+      "grad_norm": 0.5029733604515428,
+      "learning_rate": 5.094054112608758e-06,
+      "loss": 0.3645,
+      "step": 330
+    },
+    {
+      "epoch": 0.9910179640718563,
+      "grad_norm": 0.39254468493039624,
+      "learning_rate": 5.070542404770413e-06,
+      "loss": 0.3476,
+      "step": 331
+    },
+    {
+      "epoch": 0.9940119760479041,
+      "grad_norm": 0.39001311298198393,
+      "learning_rate": 5.047029136674563e-06,
+      "loss": 0.2718,
+      "step": 332
+    },
+    {
+      "epoch": 0.9970059880239521,
+      "grad_norm": 0.5023085700724956,
+      "learning_rate": 5.023514828387868e-06,
+      "loss": 0.4258,
+      "step": 333
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4161086283589412,
+      "learning_rate": 5e-06,
+      "loss": 0.294,
+      "step": 334
+    },
+    {
+      "epoch": 1.0029940119760479,
+      "grad_norm": 0.44854744635398847,
+      "learning_rate": 4.976485171612134e-06,
+      "loss": 0.2263,
+      "step": 335
+    },
+    {
+      "epoch": 1.0059880239520957,
+      "grad_norm": 0.3792600471166041,
+      "learning_rate": 4.95297086332544e-06,
+      "loss": 0.2397,
+      "step": 336
+    },
+    {
+      "epoch": 1.0089820359281436,
+      "grad_norm": 0.4461795319670102,
+      "learning_rate": 4.9294575952295896e-06,
+      "loss": 0.2576,
+      "step": 337
+    },
+    {
+      "epoch": 1.0119760479041917,
+      "grad_norm": 0.4778779008833367,
+      "learning_rate": 4.905945887391242e-06,
+      "loss": 0.2855,
+      "step": 338
+    },
+    {
+      "epoch": 1.0149700598802396,
+      "grad_norm": 0.4524301812074451,
+      "learning_rate": 4.882436259842556e-06,
+      "loss": 0.2629,
+      "step": 339
+    },
+    {
+      "epoch": 1.0179640718562875,
+      "grad_norm": 0.40236444615657746,
+      "learning_rate": 4.858929232569671e-06,
+      "loss": 0.264,
+      "step": 340
+    },
+    {
+      "epoch": 1.0209580838323353,
+      "grad_norm": 0.4159832200546139,
+      "learning_rate": 4.835425325501214e-06,
+      "loss": 0.2665,
+      "step": 341
+    },
+    {
+      "epoch": 1.0239520958083832,
+      "grad_norm": 0.3775897921199922,
+      "learning_rate": 4.811925058496799e-06,
+      "loss": 0.2131,
+      "step": 342
+    },
+    {
+      "epoch": 1.026946107784431,
+      "grad_norm": 0.44772603026013075,
+      "learning_rate": 4.788428951335534e-06,
+      "loss": 0.278,
+      "step": 343
+    },
+    {
+      "epoch": 1.029940119760479,
+      "grad_norm": 0.3684402858054329,
+      "learning_rate": 4.7649375237045135e-06,
+      "loss": 0.2488,
+      "step": 344
+    },
+    {
+      "epoch": 1.032934131736527,
+      "grad_norm": 0.3751987470575119,
+      "learning_rate": 4.741451295187333e-06,
+      "loss": 0.2219,
+      "step": 345
+    },
+    {
+      "epoch": 1.035928143712575,
+      "grad_norm": 0.3766466617292629,
+      "learning_rate": 4.717970785252596e-06,
+      "loss": 0.2952,
+      "step": 346
+    },
+    {
+      "epoch": 1.0389221556886228,
+      "grad_norm": 0.3887666015015033,
+      "learning_rate": 4.694496513242418e-06,
+      "loss": 0.2823,
+      "step": 347
+    },
+    {
+      "epoch": 1.0419161676646707,
+      "grad_norm": 0.4266354806923591,
+      "learning_rate": 4.671028998360947e-06,
+      "loss": 0.2345,
+      "step": 348
+    },
+    {
+      "epoch": 1.0449101796407185,
+      "grad_norm": 0.42274661610828324,
+      "learning_rate": 4.647568759662876e-06,
+      "loss": 0.2882,
+      "step": 349
+    },
+    {
+      "epoch": 1.0479041916167664,
+      "grad_norm": 0.40339281683858236,
+      "learning_rate": 4.624116316041962e-06,
+      "loss": 0.2709,
+      "step": 350
+    },
+    {
+      "epoch": 1.0508982035928143,
+      "grad_norm": 0.41610621967479283,
+      "learning_rate": 4.600672186219551e-06,
+      "loss": 0.2732,
+      "step": 351
+    },
+    {
+      "epoch": 1.0538922155688624,
+      "grad_norm": 0.4561758093677038,
+      "learning_rate": 4.5772368887331044e-06,
+      "loss": 0.3084,
+      "step": 352
+    },
+    {
+      "epoch": 1.0568862275449102,
+      "grad_norm": 0.41535347141162576,
+      "learning_rate": 4.553810941924735e-06,
+      "loss": 0.2989,
+      "step": 353
+    },
+    {
+      "epoch": 1.0598802395209581,
+      "grad_norm": 0.42866645714984486,
+      "learning_rate": 4.530394863929728e-06,
+      "loss": 0.2631,
+      "step": 354
+    },
+    {
+      "epoch": 1.062874251497006,
+      "grad_norm": 0.4413175613167669,
+      "learning_rate": 4.506989172665097e-06,
+      "loss": 0.2545,
+      "step": 355
+    },
+    {
+      "epoch": 1.0658682634730539,
+      "grad_norm": 0.4471826257273022,
+      "learning_rate": 4.483594385818119e-06,
+      "loss": 0.2472,
+      "step": 356
+    },
+    {
+      "epoch": 1.0688622754491017,
+      "grad_norm": 0.46865556582457474,
+      "learning_rate": 4.460211020834887e-06,
+      "loss": 0.2638,
+      "step": 357
+    },
+    {
+      "epoch": 1.0718562874251496,
+      "grad_norm": 0.47788908624330095,
+      "learning_rate": 4.436839594908866e-06,
+      "loss": 0.3281,
+      "step": 358
+    },
+    {
+      "epoch": 1.0748502994011977,
+      "grad_norm": 0.4988837776404189,
+      "learning_rate": 4.4134806249694514e-06,
+      "loss": 0.2952,
+      "step": 359
+    },
+    {
+      "epoch": 1.0778443113772456,
+      "grad_norm": 0.47850153081951347,
+      "learning_rate": 4.39013462767054e-06,
+      "loss": 0.3023,
+      "step": 360
+    },
+    {
+      "epoch": 1.0808383233532934,
+      "grad_norm": 0.40231684684878943,
+      "learning_rate": 4.366802119379098e-06,
+      "loss": 0.2063,
+      "step": 361
+    },
+    {
+      "epoch": 1.0838323353293413,
+      "grad_norm": 0.3914043084876908,
+      "learning_rate": 4.34348361616374e-06,
+      "loss": 0.3078,
+      "step": 362
+    },
+    {
+      "epoch": 1.0868263473053892,
+      "grad_norm": 0.5322637028414254,
+      "learning_rate": 4.3201796337833165e-06,
+      "loss": 0.2959,
+      "step": 363
+    },
+    {
+      "epoch": 1.089820359281437,
+      "grad_norm": 0.5396302699948718,
+      "learning_rate": 4.29689068767551e-06,
+      "loss": 0.3074,
+      "step": 364
+    },
+    {
+      "epoch": 1.092814371257485,
+      "grad_norm": 0.4327804469964399,
+      "learning_rate": 4.273617292945425e-06,
+      "loss": 0.2617,
+      "step": 365
+    },
+    {
+      "epoch": 1.095808383233533,
+      "grad_norm": 0.40367447064519496,
+      "learning_rate": 4.250359964354203e-06,
+      "loss": 0.277,
+      "step": 366
+    },
+    {
+      "epoch": 1.098802395209581,
+      "grad_norm": 0.4793349548929775,
+      "learning_rate": 4.227119216307637e-06,
+      "loss": 0.2824,
+      "step": 367
+    },
+    {
+      "epoch": 1.1017964071856288,
+      "grad_norm": 0.44071205082713466,
+      "learning_rate": 4.203895562844789e-06,
+      "loss": 0.3049,
+      "step": 368
+    },
+    {
+      "epoch": 1.1047904191616766,
+      "grad_norm": 0.487118683147286,
+      "learning_rate": 4.18068951762662e-06,
+      "loss": 0.3205,
+      "step": 369
+    },
+    {
+      "epoch": 1.1077844311377245,
+      "grad_norm": 0.41581439718297675,
+      "learning_rate": 4.157501593924638e-06,
+      "loss": 0.2445,
+      "step": 370
+    },
+    {
+      "epoch": 1.1107784431137724,
+      "grad_norm": 0.42985557719044726,
+      "learning_rate": 4.134332304609533e-06,
+      "loss": 0.2908,
+      "step": 371
+    },
+    {
+      "epoch": 1.1137724550898203,
+      "grad_norm": 0.34494159820070774,
+      "learning_rate": 4.111182162139844e-06,
+      "loss": 0.2643,
+      "step": 372
+    },
+    {
+      "epoch": 1.1167664670658684,
+      "grad_norm": 0.4536818369410819,
+      "learning_rate": 4.088051678550617e-06,
+      "loss": 0.254,
+      "step": 373
+    },
+    {
+      "epoch": 1.1197604790419162,
+      "grad_norm": 0.46241970910580693,
+      "learning_rate": 4.064941365442084e-06,
+      "loss": 0.2824,
+      "step": 374
+    },
+    {
+      "epoch": 1.122754491017964,
+      "grad_norm": 0.5296318464373709,
+      "learning_rate": 4.041851733968348e-06,
+      "loss": 0.325,
+      "step": 375
+    },
+    {
+      "epoch": 1.125748502994012,
+      "grad_norm": 0.4565857414202669,
+      "learning_rate": 4.018783294826071e-06,
+      "loss": 0.2954,
+      "step": 376
+    },
+    {
+      "epoch": 1.1287425149700598,
+      "grad_norm": 0.4563666092088836,
+      "learning_rate": 3.995736558243186e-06,
+      "loss": 0.3086,
+      "step": 377
+    },
+    {
+      "epoch": 1.1317365269461077,
+      "grad_norm": 0.42492166975578083,
+      "learning_rate": 3.972712033967608e-06,
+      "loss": 0.2763,
+      "step": 378
+    },
+    {
+      "epoch": 1.1347305389221556,
+      "grad_norm": 0.3991661977814953,
+      "learning_rate": 3.949710231255961e-06,
+      "loss": 0.3079,
+      "step": 379
+    },
+    {
+      "epoch": 1.1377245508982037,
+      "grad_norm": 0.5505533468658588,
+      "learning_rate": 3.926731658862307e-06,
+      "loss": 0.2806,
+      "step": 380
+    },
+    {
+      "epoch": 1.1407185628742516,
+      "grad_norm": 0.41384137815073285,
+      "learning_rate": 3.903776825026912e-06,
+      "loss": 0.3011,
+      "step": 381
+    },
+    {
+      "epoch": 1.1437125748502994,
+      "grad_norm": 0.42891844063935936,
+      "learning_rate": 3.8808462374649805e-06,
+      "loss": 0.2843,
+      "step": 382
+    },
+    {
+      "epoch": 1.1467065868263473,
+      "grad_norm": 0.4353979383092861,
+      "learning_rate": 3.857940403355444e-06,
+      "loss": 0.2525,
+      "step": 383
+    },
+    {
+      "epoch": 1.1497005988023952,
+      "grad_norm": 0.447137588869267,
+      "learning_rate": 3.8350598293297345e-06,
+      "loss": 0.2982,
+      "step": 384
+    },
+    {
+      "epoch": 1.152694610778443,
+      "grad_norm": 0.4358677233742794,
+      "learning_rate": 3.8122050214605822e-06,
+      "loss": 0.2532,
+      "step": 385
+    },
+    {
+      "epoch": 1.1556886227544911,
+      "grad_norm": 0.4103540118742374,
+      "learning_rate": 3.7893764852508207e-06,
+      "loss": 0.2799,
+      "step": 386
+    },
+    {
+      "epoch": 1.158682634730539,
+      "grad_norm": 0.41545016499032045,
+      "learning_rate": 3.766574725622208e-06,
+      "loss": 0.2457,
+      "step": 387
+    },
+    {
+      "epoch": 1.1616766467065869,
+      "grad_norm": 0.41843673096018297,
+      "learning_rate": 3.7438002469042567e-06,
+      "loss": 0.2691,
+      "step": 388
+    },
+    {
+      "epoch": 1.1646706586826348,
+      "grad_norm": 0.46973265653591106,
+      "learning_rate": 3.721053552823078e-06,
+      "loss": 0.2855,
+      "step": 389
+    },
+    {
+      "epoch": 1.1676646706586826,
+      "grad_norm": 0.4518723586462463,
+      "learning_rate": 3.698335146490246e-06,
+      "loss": 0.2688,
+      "step": 390
+    },
+    {
+      "epoch": 1.1706586826347305,
+      "grad_norm": 0.4994684779350072,
+      "learning_rate": 3.675645530391665e-06,
+      "loss": 0.2645,
+      "step": 391
+    },
+    {
+      "epoch": 1.1736526946107784,
+      "grad_norm": 0.44957162388074273,
+      "learning_rate": 3.652985206376455e-06,
+      "loss": 0.2469,
+      "step": 392
+    },
+    {
+      "epoch": 1.1766467065868262,
+      "grad_norm": 0.3523267176322131,
+      "learning_rate": 3.630354675645853e-06,
+      "loss": 0.2329,
+      "step": 393
+    },
+    {
+      "epoch": 1.1796407185628743,
+      "grad_norm": 0.4583974317337641,
+      "learning_rate": 3.6077544387421293e-06,
+      "loss": 0.2875,
+      "step": 394
+    },
+    {
+      "epoch": 1.1826347305389222,
+      "grad_norm": 0.47419741344878874,
+      "learning_rate": 3.5851849955375177e-06,
+      "loss": 0.2471,
+      "step": 395
+    },
+    {
+      "epoch": 1.18562874251497,
+      "grad_norm": 0.45299934075623194,
+      "learning_rate": 3.5626468452231534e-06,
+      "loss": 0.2608,
+      "step": 396
+    },
+    {
+      "epoch": 1.188622754491018,
+      "grad_norm": 0.42799113733804134,
+      "learning_rate": 3.540140486298035e-06,
+      "loss": 0.2135,
+      "step": 397
+    },
+    {
+      "epoch": 1.1916167664670658,
+      "grad_norm": 0.4028291897756234,
+      "learning_rate": 3.517666416557999e-06,
+      "loss": 0.304,
+      "step": 398
+    },
+    {
+      "epoch": 1.1946107784431137,
+      "grad_norm": 0.44219813309128264,
+      "learning_rate": 3.495225133084712e-06,
+      "loss": 0.2728,
+      "step": 399
+    },
+    {
+      "epoch": 1.1976047904191618,
+      "grad_norm": 0.41156772536944447,
+      "learning_rate": 3.472817132234669e-06,
+      "loss": 0.2916,
+      "step": 400
+    },
+    {
+      "epoch": 1.1976047904191618,
+      "eval_loss": 0.3157062530517578,
+      "eval_runtime": 2.335,
+      "eval_samples_per_second": 23.126,
+      "eval_steps_per_second": 5.996,
+      "step": 400
+    },
+    {
+      "epoch": 1.2005988023952097,
+      "grad_norm": 0.4286207646329637,
+      "learning_rate": 3.4504429096282246e-06,
+      "loss": 0.3005,
+      "step": 401
+    },
+    {
+      "epoch": 1.2035928143712575,
+      "grad_norm": 0.43394729493522544,
+      "learning_rate": 3.428102960138625e-06,
+      "loss": 0.2875,
+      "step": 402
+    },
+    {
+      "epoch": 1.2065868263473054,
+      "grad_norm": 0.4473235456133437,
+      "learning_rate": 3.405797777881059e-06,
+      "loss": 0.3097,
+      "step": 403
+    },
+    {
+      "epoch": 1.2095808383233533,
+      "grad_norm": 0.46059876083067086,
+      "learning_rate": 3.3835278562017405e-06,
+      "loss": 0.2508,
+      "step": 404
+    },
+    {
+      "epoch": 1.2125748502994012,
+      "grad_norm": 0.412361667170021,
+      "learning_rate": 3.3612936876669834e-06,
+      "loss": 0.2554,
+      "step": 405
+    },
+    {
+      "epoch": 1.215568862275449,
+      "grad_norm": 0.37746570182160316,
+      "learning_rate": 3.3390957640523147e-06,
+      "loss": 0.2528,
+      "step": 406
+    },
+    {
+      "epoch": 1.218562874251497,
+      "grad_norm": 0.40188320293028973,
+      "learning_rate": 3.3169345763315986e-06,
+      "loss": 0.2714,
+      "step": 407
+    },
+    {
+      "epoch": 1.221556886227545,
+      "grad_norm": 0.44041733273387823,
+      "learning_rate": 3.29481061466617e-06,
+      "loss": 0.2831,
+      "step": 408
+    },
+    {
+      "epoch": 1.2245508982035929,
+      "grad_norm": 0.47136797479974646,
+      "learning_rate": 3.2727243683940045e-06,
+      "loss": 0.264,
+      "step": 409
+    },
+    {
+      "epoch": 1.2275449101796407,
+      "grad_norm": 0.45247677839156836,
+      "learning_rate": 3.2506763260188824e-06,
+      "loss": 0.2952,
+      "step": 410
+    },
+    {
+      "epoch": 1.2305389221556886,
+      "grad_norm": 0.40935429460334466,
+      "learning_rate": 3.2286669751995905e-06,
+      "loss": 0.2773,
+      "step": 411
+    },
+    {
+      "epoch": 1.2335329341317365,
+      "grad_norm": 0.48415197515961755,
+      "learning_rate": 3.2066968027391377e-06,
+      "loss": 0.2812,
+      "step": 412
+    },
+    {
+      "epoch": 1.2365269461077844,
+      "grad_norm": 0.3983807068419485,
+      "learning_rate": 3.1847662945739833e-06,
+      "loss": 0.2564,
+      "step": 413
+    },
+    {
+      "epoch": 1.2395209580838324,
+      "grad_norm": 0.46043060370621,
+      "learning_rate": 3.1628759357632943e-06,
+      "loss": 0.291,
+      "step": 414
+    },
+    {
+      "epoch": 1.2425149700598803,
+      "grad_norm": 0.47265857421815705,
+      "learning_rate": 3.1410262104782086e-06,
+      "loss": 0.2488,
+      "step": 415
+    },
+    {
+      "epoch": 1.2455089820359282,
+      "grad_norm": 0.47665800755663595,
+      "learning_rate": 3.119217601991139e-06,
+      "loss": 0.2771,
+      "step": 416
+    },
+    {
+      "epoch": 1.248502994011976,
+      "grad_norm": 0.4622274618210644,
+      "learning_rate": 3.0974505926650724e-06,
+      "loss": 0.2313,
+      "step": 417
+    },
+    {
+      "epoch": 1.251497005988024,
+      "grad_norm": 0.40684217812956275,
+      "learning_rate": 3.0757256639429027e-06,
+      "loss": 0.2638,
+      "step": 418
+    },
+    {
+      "epoch": 1.2544910179640718,
+      "grad_norm": 0.4648897015164532,
+      "learning_rate": 3.0540432963367907e-06,
+      "loss": 0.3051,
+      "step": 419
+    },
+    {
+      "epoch": 1.2574850299401197,
+      "grad_norm": 0.44652170984876893,
+      "learning_rate": 3.032403969417523e-06,
+      "loss": 0.2598,
+      "step": 420
+    },
+    {
+      "epoch": 1.2604790419161676,
+      "grad_norm": 0.4110299915327591,
+      "learning_rate": 3.010808161803917e-06,
+      "loss": 0.2427,
+      "step": 421
+    },
+    {
+      "epoch": 1.2634730538922156,
+      "grad_norm": 0.4252555461819149,
+      "learning_rate": 2.9892563511522305e-06,
+      "loss": 0.3051,
+      "step": 422
+    },
+    {
+      "epoch": 1.2664670658682635,
+      "grad_norm": 0.4390072990501312,
+      "learning_rate": 2.9677490141455915e-06,
+      "loss": 0.2512,
+      "step": 423
+    },
+    {
+      "epoch": 1.2694610778443114,
+      "grad_norm": 0.43659165480416035,
+      "learning_rate": 2.946286626483463e-06,
+      "loss": 0.3007,
+      "step": 424
+    },
+    {
+      "epoch": 1.2724550898203593,
+      "grad_norm": 0.4927655018046331,
+      "learning_rate": 2.924869662871117e-06,
+      "loss": 0.2726,
+      "step": 425
+    },
+    {
+      "epoch": 1.2754491017964071,
+      "grad_norm": 0.4193484942221608,
+      "learning_rate": 2.903498597009136e-06,
+      "loss": 0.2756,
+      "step": 426
+    },
+    {
+      "epoch": 1.278443113772455,
+      "grad_norm": 0.438258443794064,
+      "learning_rate": 2.8821739015829338e-06,
+      "loss": 0.2723,
+      "step": 427
+    },
+    {
+      "epoch": 1.281437125748503,
+      "grad_norm": 0.45054239270178303,
+      "learning_rate": 2.8608960482523058e-06,
+      "loss": 0.2731,
+      "step": 428
+    },
+    {
+      "epoch": 1.284431137724551,
+      "grad_norm": 0.42821151373138516,
+      "learning_rate": 2.839665507640992e-06,
+      "loss": 0.2648,
+      "step": 429
+    },
+    {
+      "epoch": 1.2874251497005988,
+      "grad_norm": 0.4058876632612295,
+      "learning_rate": 2.818482749326272e-06,
+      "loss": 0.3111,
+      "step": 430
+    },
+    {
+      "epoch": 1.2904191616766467,
+      "grad_norm": 0.40964919970999497,
+      "learning_rate": 2.797348241828569e-06,
+      "loss": 0.2742,
+      "step": 431
+    },
+    {
+      "epoch": 1.2934131736526946,
+      "grad_norm": 0.40617738823141325,
+      "learning_rate": 2.776262452601104e-06,
+      "loss": 0.2963,
+      "step": 432
+    },
+    {
+      "epoch": 1.2964071856287425,
+      "grad_norm": 0.45274886163034556,
+      "learning_rate": 2.7552258480195348e-06,
+      "loss": 0.2767,
+      "step": 433
+    },
+    {
+      "epoch": 1.2994011976047903,
+      "grad_norm": 0.39789306683102144,
+      "learning_rate": 2.734238893371667e-06,
+      "loss": 0.2563,
+      "step": 434
+    },
+    {
+      "epoch": 1.3023952095808382,
+      "grad_norm": 0.4275855052400172,
+      "learning_rate": 2.7133020528471322e-06,
+      "loss": 0.2652,
+      "step": 435
+    },
+    {
+      "epoch": 1.3053892215568863,
+      "grad_norm": 0.40781680178405505,
+      "learning_rate": 2.6924157895271563e-06,
+      "loss": 0.2781,
+      "step": 436
+    },
+    {
+      "epoch": 1.3083832335329342,
+      "grad_norm": 0.4333699596171378,
+      "learning_rate": 2.671580565374282e-06,
+      "loss": 0.3015,
+      "step": 437
+    },
+    {
+      "epoch": 1.311377245508982,
+      "grad_norm": 0.43615293492930635,
+      "learning_rate": 2.6507968412221763e-06,
+      "loss": 0.2332,
+      "step": 438
+    },
+    {
+      "epoch": 1.31437125748503,
+      "grad_norm": 0.48560365713843645,
+      "learning_rate": 2.6300650767654234e-06,
+      "loss": 0.2646,
+      "step": 439
+    },
+    {
+      "epoch": 1.3173652694610778,
+      "grad_norm": 0.4101309718009534,
+      "learning_rate": 2.6093857305493666e-06,
+      "loss": 0.2717,
+      "step": 440
+    },
+    {
+      "epoch": 1.3203592814371259,
+      "grad_norm": 0.43252181660316574,
+      "learning_rate": 2.588759259959962e-06,
+      "loss": 0.2542,
+      "step": 441
+    },
+    {
+      "epoch": 1.3233532934131738,
+      "grad_norm": 0.4213321977668779,
+      "learning_rate": 2.568186121213658e-06,
+      "loss": 0.2868,
+      "step": 442
+    },
+    {
+      "epoch": 1.3263473053892216,
+      "grad_norm": 0.4426026801559286,
+      "learning_rate": 2.547666769347312e-06,
+      "loss": 0.2885,
+      "step": 443
+    },
+    {
+      "epoch": 1.3293413173652695,
+      "grad_norm": 0.43011783809750526,
+      "learning_rate": 2.5272016582081236e-06,
+      "loss": 0.3325,
+      "step": 444
+    },
+    {
+      "epoch": 1.3323353293413174,
+      "grad_norm": 0.44842448774324506,
+      "learning_rate": 2.5067912404435952e-06,
+      "loss": 0.2863,
+      "step": 445
+    },
+    {
+      "epoch": 1.3353293413173652,
+      "grad_norm": 0.43828674670735285,
+      "learning_rate": 2.486435967491516e-06,
+      "loss": 0.2596,
+      "step": 446
+    },
+    {
+      "epoch": 1.3383233532934131,
+      "grad_norm": 0.4110650547870287,
+      "learning_rate": 2.4661362895699903e-06,
+      "loss": 0.2537,
+      "step": 447
+    },
+    {
+      "epoch": 1.341317365269461,
+      "grad_norm": 0.4572894177808882,
+      "learning_rate": 2.445892655667462e-06,
+      "loss": 0.344,
+      "step": 448
+    },
+    {
+      "epoch": 1.3443113772455089,
+      "grad_norm": 0.47126513900649314,
+      "learning_rate": 2.425705513532798e-06,
+      "loss": 0.295,
+      "step": 449
+    },
+    {
+      "epoch": 1.347305389221557,
+      "grad_norm": 0.4121224940282215,
+      "learning_rate": 2.4055753096653795e-06,
+      "loss": 0.2942,
+      "step": 450
+    },
+    {
+      "epoch": 1.3502994011976048,
+      "grad_norm": 0.3930595163215845,
+      "learning_rate": 2.3855024893052286e-06,
+      "loss": 0.2413,
+      "step": 451
+    },
+    {
+      "epoch": 1.3532934131736527,
+      "grad_norm": 0.43263302405036524,
+      "learning_rate": 2.365487496423152e-06,
+      "loss": 0.2682,
+      "step": 452
+    },
+    {
+      "epoch": 1.3562874251497006,
+      "grad_norm": 0.39335461108416525,
+      "learning_rate": 2.3455307737109338e-06,
+      "loss": 0.2318,
+      "step": 453
+    },
+    {
+      "epoch": 1.3592814371257484,
+      "grad_norm": 0.410165917949307,
+      "learning_rate": 2.3256327625715345e-06,
+      "loss": 0.2344,
+      "step": 454
+    },
+    {
+      "epoch": 1.3622754491017965,
+      "grad_norm": 0.45976619638762756,
+      "learning_rate": 2.3057939031093346e-06,
+      "loss": 0.2666,
+      "step": 455
+    },
+    {
+      "epoch": 1.3652694610778444,
+      "grad_norm": 0.3915630610257879,
+      "learning_rate": 2.2860146341203936e-06,
+      "loss": 0.2559,
+      "step": 456
+    },
+    {
+      "epoch": 1.3682634730538923,
+      "grad_norm": 0.4554169379145131,
+      "learning_rate": 2.2662953930827546e-06,
+      "loss": 0.3312,
+      "step": 457
+    },
+    {
+      "epoch": 1.3712574850299402,
+      "grad_norm": 0.45984848993066124,
+      "learning_rate": 2.2466366161467528e-06,
+      "loss": 0.2684,
+      "step": 458
+    },
+    {
+      "epoch": 1.374251497005988,
+      "grad_norm": 0.4812754060619508,
+      "learning_rate": 2.227038738125385e-06,
+      "loss": 0.2497,
+      "step": 459
+    },
+    {
+      "epoch": 1.377245508982036,
+      "grad_norm": 0.4958879650824535,
+      "learning_rate": 2.207502192484685e-06,
+      "loss": 0.266,
+      "step": 460
+    },
+    {
+      "epoch": 1.3802395209580838,
+      "grad_norm": 0.4142114799645694,
+      "learning_rate": 2.188027411334131e-06,
+      "loss": 0.2667,
+      "step": 461
+    },
+    {
+      "epoch": 1.3832335329341316,
+      "grad_norm": 0.47456645321918406,
+      "learning_rate": 2.1686148254171012e-06,
+      "loss": 0.2651,
+      "step": 462
+    },
+    {
+      "epoch": 1.3862275449101795,
+      "grad_norm": 0.5255584756587222,
+      "learning_rate": 2.1492648641013305e-06,
+      "loss": 0.2666,
+      "step": 463
+    },
+    {
+      "epoch": 1.3892215568862276,
+      "grad_norm": 0.3821200807234482,
+      "learning_rate": 2.1299779553694323e-06,
+      "loss": 0.2433,
+      "step": 464
+    },
+    {
+      "epoch": 1.3922155688622755,
+      "grad_norm": 0.48216854818315175,
+      "learning_rate": 2.1107545258094135e-06,
+      "loss": 0.2701,
+      "step": 465
+    },
+    {
+      "epoch": 1.3952095808383234,
+      "grad_norm": 0.44623157356945264,
+      "learning_rate": 2.0915950006052555e-06,
+      "loss": 0.2288,
+      "step": 466
+    },
+    {
+      "epoch": 1.3982035928143712,
+      "grad_norm": 0.3946784024752767,
+      "learning_rate": 2.0724998035274947e-06,
+      "loss": 0.2507,
+      "step": 467
+    },
+    {
+      "epoch": 1.401197604790419,
+      "grad_norm": 0.4439008726980056,
+      "learning_rate": 2.053469356923865e-06,
+      "loss": 0.2236,
+      "step": 468
+    },
+    {
+      "epoch": 1.4041916167664672,
+      "grad_norm": 0.43254035615679426,
+      "learning_rate": 2.0345040817099433e-06,
+      "loss": 0.2705,
+      "step": 469
+    },
+    {
+      "epoch": 1.407185628742515,
+      "grad_norm": 0.3950126318223767,
+      "learning_rate": 2.0156043973598475e-06,
+      "loss": 0.2481,
+      "step": 470
+    },
+    {
+      "epoch": 1.410179640718563,
+      "grad_norm": 0.47678450235241576,
+      "learning_rate": 1.996770721896957e-06,
+      "loss": 0.2801,
+      "step": 471
+    },
+    {
+      "epoch": 1.4131736526946108,
+      "grad_norm": 0.48436482686808946,
+      "learning_rate": 1.9780034718846653e-06,
+      "loss": 0.2702,
+      "step": 472
+    },
+    {
+      "epoch": 1.4161676646706587,
+      "grad_norm": 0.41763409142515995,
+      "learning_rate": 1.9593030624171683e-06,
+      "loss": 0.2926,
+      "step": 473
+    },
+    {
+      "epoch": 1.4191616766467066,
+      "grad_norm": 0.46350324380929653,
+      "learning_rate": 1.9406699071102774e-06,
+      "loss": 0.3009,
+      "step": 474
+    },
+    {
+      "epoch": 1.4221556886227544,
+      "grad_norm": 0.4205958530266388,
+      "learning_rate": 1.9221044180922833e-06,
+      "loss": 0.3037,
+      "step": 475
+    },
+    {
+      "epoch": 1.4251497005988023,
+      "grad_norm": 0.4250636865469936,
+      "learning_rate": 1.9036070059948253e-06,
+      "loss": 0.2687,
+      "step": 476
+    },
+    {
+      "epoch": 1.4281437125748502,
+      "grad_norm": 0.4918933827960068,
+      "learning_rate": 1.885178079943823e-06,
+      "loss": 0.3028,
+      "step": 477
+    },
+    {
+      "epoch": 1.4311377245508983,
+      "grad_norm": 0.4511636427629257,
+      "learning_rate": 1.866818047550419e-06,
+      "loss": 0.3442,
+      "step": 478
+    },
+    {
+      "epoch": 1.4341317365269461,
+      "grad_norm": 0.4115103054450131,
+      "learning_rate": 1.8485273149019655e-06,
+      "loss": 0.2635,
+      "step": 479
+    },
+    {
+      "epoch": 1.437125748502994,
+      "grad_norm": 0.44813899639097676,
+      "learning_rate": 1.8303062865530407e-06,
+      "loss": 0.247,
+      "step": 480
+    },
+    {
+      "epoch": 1.4401197604790419,
+      "grad_norm": 0.4418545172687306,
+      "learning_rate": 1.8121553655165058e-06,
+      "loss": 0.2547,
+      "step": 481
+    },
+    {
+      "epoch": 1.4431137724550898,
+      "grad_norm": 0.4569720439493011,
+      "learning_rate": 1.7940749532545832e-06,
+      "loss": 0.3109,
+      "step": 482
+    },
+    {
+      "epoch": 1.4461077844311379,
+      "grad_norm": 0.5076970735363964,
+      "learning_rate": 1.7760654496699876e-06,
+      "loss": 0.269,
+      "step": 483
+    },
+    {
+      "epoch": 1.4491017964071857,
+      "grad_norm": 0.4375401332723699,
+      "learning_rate": 1.7581272530970666e-06,
+      "loss": 0.2838,
+      "step": 484
+    },
+    {
+      "epoch": 1.4520958083832336,
+      "grad_norm": 0.3992861416112489,
+      "learning_rate": 1.7402607602930106e-06,
+      "loss": 0.2611,
+      "step": 485
+    },
+    {
+      "epoch": 1.4550898203592815,
+      "grad_norm": 0.48989543357544685,
+      "learning_rate": 1.7224663664290537e-06,
+      "loss": 0.327,
+      "step": 486
+    },
+    {
+      "epoch": 1.4580838323353293,
+      "grad_norm": 0.44124257933949074,
+      "learning_rate": 1.7047444650817518e-06,
+      "loss": 0.2697,
+      "step": 487
+    },
+    {
+      "epoch": 1.4610778443113772,
+      "grad_norm": 0.4838831046559625,
+      "learning_rate": 1.6870954482242707e-06,
+      "loss": 0.254,
+      "step": 488
+    },
+    {
+      "epoch": 1.464071856287425,
+      "grad_norm": 0.41761518020594485,
+      "learning_rate": 1.669519706217711e-06,
+      "loss": 0.2337,
+      "step": 489
+    },
+    {
+      "epoch": 1.467065868263473,
+      "grad_norm": 0.4480813110262778,
+      "learning_rate": 1.652017627802487e-06,
+      "loss": 0.2836,
+      "step": 490
+    },
+    {
+      "epoch": 1.4700598802395208,
+      "grad_norm": 0.39311060253238145,
+      "learning_rate": 1.6345896000897122e-06,
+      "loss": 0.2455,
+      "step": 491
+    },
+    {
+      "epoch": 1.473053892215569,
+      "grad_norm": 0.3858430272701863,
+      "learning_rate": 1.6172360085526567e-06,
+      "loss": 0.2834,
+      "step": 492
+    },
+    {
+      "epoch": 1.4760479041916168,
+      "grad_norm": 0.377972083418802,
+      "learning_rate": 1.5999572370182016e-06,
+      "loss": 0.2777,
+      "step": 493
+    },
+    {
+      "epoch": 1.4790419161676647,
+      "grad_norm": 0.46367794989138517,
+      "learning_rate": 1.5827536676583643e-06,
+      "loss": 0.2847,
+      "step": 494
+    },
+    {
+      "epoch": 1.4820359281437125,
+      "grad_norm": 0.431668507061833,
+      "learning_rate": 1.5656256809818343e-06,
+      "loss": 0.2561,
+      "step": 495
+    },
+    {
+      "epoch": 1.4850299401197604,
+      "grad_norm": 0.46268719385530516,
+      "learning_rate": 1.54857365582557e-06,
+      "loss": 0.2313,
+      "step": 496
+    },
+    {
+      "epoch": 1.4880239520958085,
+      "grad_norm": 0.43552214063565425,
+      "learning_rate": 1.5315979693464039e-06,
+      "loss": 0.2819,
+      "step": 497
+    },
+    {
+      "epoch": 1.4910179640718564,
+      "grad_norm": 0.4943347391827864,
+      "learning_rate": 1.5146989970127158e-06,
+      "loss": 0.2346,
+      "step": 498
+    },
+    {
+      "epoch": 1.4940119760479043,
+      "grad_norm": 0.45110132647814694,
+      "learning_rate": 1.4978771125961177e-06,
+      "loss": 0.2756,
+      "step": 499
+    },
+    {
+      "epoch": 1.4970059880239521,
+      "grad_norm": 0.392235386713459,
+      "learning_rate": 1.4811326881631937e-06,
+      "loss": 0.274,
+      "step": 500
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.42985594662811255,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.2744,
+      "step": 501
+    },
+    {
+      "epoch": 1.5029940119760479,
+      "grad_norm": 0.45080263305728446,
+      "learning_rate": 1.4478776989401949e-06,
+      "loss": 0.2905,
+      "step": 502
+    },
+    {
+      "epoch": 1.5059880239520957,
+      "grad_norm": 0.45840668881918484,
+      "learning_rate": 1.4313678696842559e-06,
+      "loss": 0.2637,
+      "step": 503
+    },
+    {
+      "epoch": 1.5089820359281436,
+      "grad_norm": 0.4322855428598552,
+      "learning_rate": 1.4149369714639856e-06,
+      "loss": 0.2598,
+      "step": 504
+    },
+    {
+      "epoch": 1.5119760479041915,
+      "grad_norm": 0.44134762297622204,
+      "learning_rate": 1.3985853676981316e-06,
+      "loss": 0.2691,
+      "step": 505
+    },
+    {
+      "epoch": 1.5149700598802394,
+      "grad_norm": 0.45963375219819363,
+      "learning_rate": 1.3823134200516043e-06,
+      "loss": 0.265,
+      "step": 506
+    },
+    {
+      "epoch": 1.5179640718562875,
+      "grad_norm": 0.4929631398468762,
+      "learning_rate": 1.366121488427481e-06,
+      "loss": 0.2676,
+      "step": 507
+    },
+    {
+      "epoch": 1.5209580838323353,
+      "grad_norm": 0.47565138080230834,
+      "learning_rate": 1.3500099309590397e-06,
+      "loss": 0.2813,
+      "step": 508
+    },
+    {
+      "epoch": 1.5239520958083832,
+      "grad_norm": 0.4686523800649796,
+      "learning_rate": 1.3339791040018479e-06,
+      "loss": 0.3143,
+      "step": 509
+    },
+    {
+      "epoch": 1.5269461077844313,
+      "grad_norm": 0.40534835902619354,
+      "learning_rate": 1.3180293621258694e-06,
+      "loss": 0.2696,
+      "step": 510
+    },
+    {
+      "epoch": 1.5299401197604792,
+      "grad_norm": 0.45650700865913724,
+      "learning_rate": 1.3021610581076316e-06,
+      "loss": 0.2362,
+      "step": 511
+    },
+    {
+      "epoch": 1.532934131736527,
+      "grad_norm": 0.49309443069007575,
+      "learning_rate": 1.2863745429224145e-06,
+      "loss": 0.2736,
+      "step": 512
+    },
+    {
+      "epoch": 1.535928143712575,
+      "grad_norm": 0.43526374834732373,
+      "learning_rate": 1.270670165736499e-06,
+      "loss": 0.2792,
+      "step": 513
+    },
+    {
+      "epoch": 1.5389221556886228,
+      "grad_norm": 0.5269060701254544,
+      "learning_rate": 1.2550482738994284e-06,
+      "loss": 0.3261,
+      "step": 514
+    },
+    {
+      "epoch": 1.5419161676646707,
+      "grad_norm": 0.4263070903322032,
+      "learning_rate": 1.239509212936343e-06,
+      "loss": 0.2678,
+      "step": 515
+    },
+    {
+      "epoch": 1.5449101796407185,
+      "grad_norm": 0.44924951220050646,
+      "learning_rate": 1.22405332654032e-06,
+      "loss": 0.2545,
+      "step": 516
+    },
+    {
+      "epoch": 1.5479041916167664,
+      "grad_norm": 0.3796212997364053,
+      "learning_rate": 1.2086809565647877e-06,
+      "loss": 0.2495,
+      "step": 517
+    },
+    {
+      "epoch": 1.5508982035928143,
+      "grad_norm": 0.44641467989586464,
+      "learning_rate": 1.1933924430159571e-06,
+      "loss": 0.2679,
+      "step": 518
+    },
+    {
+      "epoch": 1.5538922155688621,
+      "grad_norm": 0.4456331064818211,
+      "learning_rate": 1.1781881240452958e-06,
+      "loss": 0.2707,
+      "step": 519
+    },
+    {
+      "epoch": 1.55688622754491,
+      "grad_norm": 0.44656888036518866,
+      "learning_rate": 1.1630683359420653e-06,
+      "loss": 0.2697,
+      "step": 520
+    },
+    {
+      "epoch": 1.5598802395209581,
+      "grad_norm": 0.4599470832880608,
+      "learning_rate": 1.1480334131258626e-06,
+      "loss": 0.2182,
+      "step": 521
+    },
+    {
+      "epoch": 1.562874251497006,
+      "grad_norm": 0.4777818849135313,
+      "learning_rate": 1.1330836881392405e-06,
+      "loss": 0.2578,
+      "step": 522
+    },
+    {
+      "epoch": 1.5658682634730539,
+      "grad_norm": 0.45240864219805954,
+      "learning_rate": 1.11821949164034e-06,
+      "loss": 0.2846,
+      "step": 523
+    },
+    {
+      "epoch": 1.568862275449102,
+      "grad_norm": 0.5108752456653508,
+      "learning_rate": 1.103441152395588e-06,
+      "loss": 0.3281,
+      "step": 524
+    },
+    {
+      "epoch": 1.5718562874251498,
+      "grad_norm": 0.46383753493242447,
+      "learning_rate": 1.088748997272414e-06,
+      "loss": 0.2878,
+      "step": 525
+    },
+    {
+      "epoch": 1.5748502994011977,
+      "grad_norm": 0.42919867400510703,
+      "learning_rate": 1.0741433512320316e-06,
+      "loss": 0.2217,
+      "step": 526
+    },
+    {
+      "epoch": 1.5778443113772456,
+      "grad_norm": 0.498020433144664,
+      "learning_rate": 1.0596245373222424e-06,
+      "loss": 0.2519,
+      "step": 527
+    },
+    {
+      "epoch": 1.5808383233532934,
+      "grad_norm": 0.44075473198085663,
+      "learning_rate": 1.045192876670298e-06,
+      "loss": 0.2975,
+      "step": 528
+    },
+    {
+      "epoch": 1.5838323353293413,
+      "grad_norm": 0.43625916956401495,
+      "learning_rate": 1.0308486884757868e-06,
+      "loss": 0.3013,
+      "step": 529
+    },
+    {
+      "epoch": 1.5868263473053892,
+      "grad_norm": 0.39003496624685086,
+      "learning_rate": 1.0165922900035886e-06,
+      "loss": 0.2706,
+      "step": 530
+    },
+    {
+      "epoch": 1.589820359281437,
+      "grad_norm": 0.4313983939608606,
+      "learning_rate": 1.0024239965768417e-06,
+      "loss": 0.26,
+      "step": 531
+    },
+    {
+      "epoch": 1.592814371257485,
+      "grad_norm": 0.46461251183175495,
+      "learning_rate": 9.883441215699824e-07,
+      "loss": 0.2073,
+      "step": 532
+    },
+    {
+      "epoch": 1.5958083832335328,
+      "grad_norm": 0.525799801366948,
+      "learning_rate": 9.74352976401805e-07,
+      "loss": 0.3286,
+      "step": 533
+    },
+    {
+      "epoch": 1.5988023952095807,
+      "grad_norm": 0.489125166189841,
+      "learning_rate": 9.604508705285765e-07,
+      "loss": 0.2734,
+      "step": 534
+    },
+    {
+      "epoch": 1.6017964071856288,
+      "grad_norm": 0.46489046995310773,
+      "learning_rate": 9.466381114371942e-07,
+      "loss": 0.2741,
+      "step": 535
+    },
+    {
+      "epoch": 1.6047904191616766,
+      "grad_norm": 0.445456392161074,
+      "learning_rate": 9.329150046383773e-07,
+      "loss": 0.2615,
+      "step": 536
+    },
+    {
+      "epoch": 1.6077844311377245,
+      "grad_norm": 0.466584318565915,
+      "learning_rate": 9.192818536599213e-07,
+      "loss": 0.2718,
+      "step": 537
+    },
+    {
+      "epoch": 1.6107784431137726,
+      "grad_norm": 0.43230649836687324,
+      "learning_rate": 9.057389600399719e-07,
+      "loss": 0.2764,
+      "step": 538
+    },
+    {
+      "epoch": 1.6137724550898205,
+      "grad_norm": 0.4163888213991036,
+      "learning_rate": 8.922866233203681e-07,
+      "loss": 0.256,
+      "step": 539
+    },
+    {
+      "epoch": 1.6167664670658684,
+      "grad_norm": 0.3564682935335293,
+      "learning_rate": 8.789251410400024e-07,
+      "loss": 0.2042,
+      "step": 540
+    },
+    {
+      "epoch": 1.6197604790419162,
+      "grad_norm": 0.3578811300249587,
+      "learning_rate": 8.65654808728259e-07,
+      "loss": 0.2587,
+      "step": 541
+    },
+    {
+      "epoch": 1.622754491017964,
+      "grad_norm": 0.5379787442171552,
+      "learning_rate": 8.524759198984567e-07,
+      "loss": 0.2484,
+      "step": 542
+    },
+    {
+      "epoch": 1.625748502994012,
+      "grad_norm": 0.4369271997539601,
+      "learning_rate": 8.393887660413719e-07,
+      "loss": 0.2632,
+      "step": 543
+    },
+    {
+      "epoch": 1.6287425149700598,
+      "grad_norm": 0.537604207190242,
+      "learning_rate": 8.263936366187825e-07,
+      "loss": 0.3165,
+      "step": 544
+    },
+    {
+      "epoch": 1.6317365269461077,
+      "grad_norm": 0.4241497298865533,
+      "learning_rate": 8.134908190570723e-07,
+      "loss": 0.2913,
+      "step": 545
+    },
+    {
+      "epoch": 1.6347305389221556,
+      "grad_norm": 0.3860958910004761,
+      "learning_rate": 8.006805987408705e-07,
+      "loss": 0.2028,
+      "step": 546
+    },
+    {
+      "epoch": 1.6377245508982035,
+      "grad_norm": 0.4641730602284422,
+      "learning_rate": 7.879632590067354e-07,
+      "loss": 0.28,
+      "step": 547
+    },
+    {
+      "epoch": 1.6407185628742516,
+      "grad_norm": 0.49251467189743703,
+      "learning_rate": 7.753390811368972e-07,
+      "loss": 0.3062,
+      "step": 548
+    },
+    {
+      "epoch": 1.6437125748502994,
+      "grad_norm": 0.4359000222635044,
+      "learning_rate": 7.628083443530287e-07,
+      "loss": 0.2662,
+      "step": 549
+    },
+    {
+      "epoch": 1.6467065868263473,
+      "grad_norm": 0.5311550454168001,
+      "learning_rate": 7.503713258100726e-07,
+      "loss": 0.2999,
+      "step": 550
+    },
+    {
+      "epoch": 1.6497005988023952,
+      "grad_norm": 0.4358232206467687,
+      "learning_rate": 7.380283005901084e-07,
+      "loss": 0.2764,
+      "step": 551
+    },
+    {
+      "epoch": 1.6526946107784433,
+      "grad_norm": 0.40685867325642805,
+      "learning_rate": 7.257795416962754e-07,
+      "loss": 0.2468,
+      "step": 552
+    },
+    {
+      "epoch": 1.6556886227544911,
+      "grad_norm": 0.43973692133551856,
+      "learning_rate": 7.136253200467231e-07,
+      "loss": 0.2699,
+      "step": 553
+    },
+    {
+      "epoch": 1.658682634730539,
+      "grad_norm": 0.45846292278632356,
+      "learning_rate": 7.015659044686307e-07,
+      "loss": 0.2634,
+      "step": 554
+    },
+    {
+      "epoch": 1.6616766467065869,
+      "grad_norm": 0.4585254753705944,
+      "learning_rate": 6.896015616922535e-07,
+      "loss": 0.2743,
+      "step": 555
+    },
+    {
+      "epoch": 1.6646706586826348,
+      "grad_norm": 0.41919608814279025,
+      "learning_rate": 6.777325563450282e-07,
+      "loss": 0.3097,
+      "step": 556
+    },
+    {
+      "epoch": 1.6676646706586826,
+      "grad_norm": 0.42382248550471446,
+      "learning_rate": 6.659591509457125e-07,
+      "loss": 0.2579,
+      "step": 557
+    },
+    {
+      "epoch": 1.6706586826347305,
+      "grad_norm": 0.48589163005701475,
+      "learning_rate": 6.542816058985896e-07,
+      "loss": 0.2832,
+      "step": 558
+    },
+    {
+      "epoch": 1.6736526946107784,
+      "grad_norm": 0.4690814897988454,
+      "learning_rate": 6.427001794876974e-07,
+      "loss": 0.2656,
+      "step": 559
+    },
+    {
+      "epoch": 1.6766467065868262,
+      "grad_norm": 0.4068231179324803,
+      "learning_rate": 6.312151278711237e-07,
+      "loss": 0.2821,
+      "step": 560
+    },
+    {
+      "epoch": 1.6796407185628741,
+      "grad_norm": 0.4082484695149451,
+      "learning_rate": 6.198267050753387e-07,
+      "loss": 0.2677,
+      "step": 561
+    },
+    {
+      "epoch": 1.6826347305389222,
+      "grad_norm": 0.44367085134929773,
+      "learning_rate": 6.085351629895736e-07,
+      "loss": 0.2639,
+      "step": 562
+    },
+    {
+      "epoch": 1.68562874251497,
+      "grad_norm": 0.4828261984995851,
+      "learning_rate": 5.973407513602514e-07,
+      "loss": 0.2464,
+      "step": 563
+    },
+    {
+      "epoch": 1.688622754491018,
+      "grad_norm": 0.4585954199032957,
+      "learning_rate": 5.862437177854629e-07,
+      "loss": 0.2841,
+      "step": 564
+    },
+    {
+      "epoch": 1.6916167664670658,
+      "grad_norm": 0.5066883798123198,
+      "learning_rate": 5.752443077094927e-07,
+      "loss": 0.2404,
+      "step": 565
+    },
+    {
+      "epoch": 1.694610778443114,
+      "grad_norm": 0.45642544737915,
+      "learning_rate": 5.643427644173838e-07,
+      "loss": 0.2744,
+      "step": 566
+    },
+    {
+      "epoch": 1.6976047904191618,
+      "grad_norm": 0.39668606526826317,
+      "learning_rate": 5.535393290295643e-07,
+      "loss": 0.2522,
+      "step": 567
+    },
+    {
+      "epoch": 1.7005988023952097,
+      "grad_norm": 0.45961428512108454,
+      "learning_rate": 5.428342404965076e-07,
+      "loss": 0.2536,
+      "step": 568
+    },
+    {
+      "epoch": 1.7035928143712575,
+      "grad_norm": 0.4594566557519259,
+      "learning_rate": 5.322277355934557e-07,
+      "loss": 0.3061,
+      "step": 569
+    },
+    {
+      "epoch": 1.7065868263473054,
+      "grad_norm": 0.4296370360627938,
+      "learning_rate": 5.217200489151714e-07,
+      "loss": 0.2772,
+      "step": 570
+    },
+    {
+      "epoch": 1.7095808383233533,
+      "grad_norm": 0.3966423265530384,
+      "learning_rate": 5.113114128707592e-07,
+      "loss": 0.2217,
+      "step": 571
+    },
+    {
+      "epoch": 1.7125748502994012,
+      "grad_norm": 0.4569282216488069,
+      "learning_rate": 5.010020576785174e-07,
+      "loss": 0.2935,
+      "step": 572
+    },
+    {
+      "epoch": 1.715568862275449,
+      "grad_norm": 0.39757728701434486,
+      "learning_rate": 4.907922113608532e-07,
+      "loss": 0.2189,
+      "step": 573
+    },
+    {
+      "epoch": 1.718562874251497,
+      "grad_norm": 0.4514824159709257,
+      "learning_rate": 4.806820997392325e-07,
+      "loss": 0.2679,
+      "step": 574
+    },
+    {
+      "epoch": 1.7215568862275448,
+      "grad_norm": 0.4102445652303537,
+      "learning_rate": 4.7067194642919036e-07,
+      "loss": 0.2672,
+      "step": 575
+    },
+    {
+      "epoch": 1.7245508982035929,
+      "grad_norm": 0.4800196138311196,
+      "learning_rate": 4.607619728353818e-07,
+      "loss": 0.2573,
+      "step": 576
+    },
+    {
+      "epoch": 1.7275449101796407,
+      "grad_norm": 0.3988346666095939,
+      "learning_rate": 4.50952398146689e-07,
+      "loss": 0.2619,
+      "step": 577
+    },
+    {
+      "epoch": 1.7305389221556886,
+      "grad_norm": 0.4285643527335388,
+      "learning_rate": 4.4124343933136525e-07,
+      "loss": 0.2593,
+      "step": 578
+    },
+    {
+      "epoch": 1.7335329341317365,
+      "grad_norm": 0.3966568167275382,
+      "learning_rate": 4.3163531113224466e-07,
+      "loss": 0.2327,
+      "step": 579
+    },
+    {
+      "epoch": 1.7365269461077846,
+      "grad_norm": 0.45668755760161844,
+      "learning_rate": 4.221282260619891e-07,
+      "loss": 0.2773,
+      "step": 580
+    },
+    {
+      "epoch": 1.7395209580838324,
+      "grad_norm": 0.3909442242611101,
+      "learning_rate": 4.127223943983849e-07,
+      "loss": 0.2491,
+      "step": 581
+    },
+    {
+      "epoch": 1.7425149700598803,
+      "grad_norm": 0.4162201965132521,
+      "learning_rate": 4.03418024179697e-07,
+      "loss": 0.2527,
+      "step": 582
+    },
+    {
+      "epoch": 1.7455089820359282,
+      "grad_norm": 0.40812685189845693,
+      "learning_rate": 3.9421532120006544e-07,
+      "loss": 0.2377,
+      "step": 583
+    },
+    {
+      "epoch": 1.748502994011976,
+      "grad_norm": 0.41576314380636137,
+      "learning_rate": 3.851144890049535e-07,
+      "loss": 0.2585,
+      "step": 584
+    },
+    {
+      "epoch": 1.751497005988024,
+      "grad_norm": 0.46474037183810074,
+      "learning_rate": 3.761157288866418e-07,
+      "loss": 0.3128,
+      "step": 585
+    },
+    {
+      "epoch": 1.7544910179640718,
+      "grad_norm": 0.4346952453521228,
+      "learning_rate": 3.672192398797858e-07,
+      "loss": 0.2636,
+      "step": 586
+    },
+    {
+      "epoch": 1.7574850299401197,
+      "grad_norm": 0.4620305959389057,
+      "learning_rate": 3.58425218757002e-07,
+      "loss": 0.2552,
+      "step": 587
+    },
+    {
+      "epoch": 1.7604790419161676,
+      "grad_norm": 0.44439932201531757,
+      "learning_rate": 3.497338600245254e-07,
+      "loss": 0.2588,
+      "step": 588
+    },
+    {
+      "epoch": 1.7634730538922154,
+      "grad_norm": 0.42551746486047903,
+      "learning_rate": 3.4114535591790233e-07,
+      "loss": 0.2452,
+      "step": 589
+    },
+    {
+      "epoch": 1.7664670658682635,
+      "grad_norm": 0.4716244453227082,
+      "learning_rate": 3.326598963977395e-07,
+      "loss": 0.2508,
+      "step": 590
+    },
+    {
+      "epoch": 1.7694610778443114,
+      "grad_norm": 0.43631552473867063,
+      "learning_rate": 3.242776691455013e-07,
+      "loss": 0.2831,
+      "step": 591
+    },
+    {
+      "epoch": 1.7724550898203593,
+      "grad_norm": 0.34778343440590953,
+      "learning_rate": 3.159988595593616e-07,
+      "loss": 0.2285,
+      "step": 592
+    },
+    {
+      "epoch": 1.7754491017964071,
+      "grad_norm": 0.4134254784812623,
+      "learning_rate": 3.078236507501015e-07,
+      "loss": 0.229,
+      "step": 593
+    },
+    {
+      "epoch": 1.7784431137724552,
+      "grad_norm": 0.42288972849358863,
+      "learning_rate": 2.9975222353705757e-07,
+      "loss": 0.2826,
+      "step": 594
+    },
+    {
+      "epoch": 1.781437125748503,
+      "grad_norm": 0.4236167251007715,
+      "learning_rate": 2.917847564441256e-07,
+      "loss": 0.2957,
+      "step": 595
+    },
+    {
+      "epoch": 1.784431137724551,
+      "grad_norm": 0.4455898487436191,
+      "learning_rate": 2.839214256958106e-07,
+      "loss": 0.2652,
+      "step": 596
+    },
+    {
+      "epoch": 1.7874251497005988,
+      "grad_norm": 0.4467907006727813,
+      "learning_rate": 2.7616240521332884e-07,
+      "loss": 0.2483,
+      "step": 597
+    },
+    {
+      "epoch": 1.7904191616766467,
+      "grad_norm": 0.4000333448403913,
+      "learning_rate": 2.6850786661076047e-07,
+      "loss": 0.2706,
+      "step": 598
+    },
+    {
+      "epoch": 1.7934131736526946,
+      "grad_norm": 0.43762029266727503,
+      "learning_rate": 2.6095797919125533e-07,
+      "loss": 0.2517,
+      "step": 599
+    },
+    {
+      "epoch": 1.7964071856287425,
+      "grad_norm": 0.42139343636519444,
+      "learning_rate": 2.5351290994328703e-07,
+      "loss": 0.2803,
+      "step": 600
+    },
+    {
+      "epoch": 1.7964071856287425,
+      "eval_loss": 0.31208279728889465,
+      "eval_runtime": 2.3306,
+      "eval_samples_per_second": 23.17,
+      "eval_steps_per_second": 6.007,
+      "step": 600
+    },
+    {
+      "epoch": 1.7994011976047903,
+      "grad_norm": 0.3925207785103808,
+      "learning_rate": 2.4617282353696093e-07,
+      "loss": 0.2398,
+      "step": 601
+    },
+    {
+      "epoch": 1.8023952095808382,
+      "grad_norm": 0.5012486960669419,
+      "learning_rate": 2.3893788232036807e-07,
+      "loss": 0.3345,
+      "step": 602
+    },
+    {
+      "epoch": 1.805389221556886,
+      "grad_norm": 0.43005153053481016,
+      "learning_rate": 2.318082463160032e-07,
+      "loss": 0.2956,
+      "step": 603
+    },
+    {
+      "epoch": 1.8083832335329342,
+      "grad_norm": 0.37002127130839935,
+      "learning_rate": 2.2478407321721295e-07,
+      "loss": 0.2455,
+      "step": 604
+    },
+    {
+      "epoch": 1.811377245508982,
+      "grad_norm": 0.3970781425566954,
+      "learning_rate": 2.1786551838471892e-07,
+      "loss": 0.257,
+      "step": 605
+    },
+    {
+      "epoch": 1.81437125748503,
+      "grad_norm": 0.47328482460022514,
+      "learning_rate": 2.1105273484317402e-07,
+      "loss": 0.2221,
+      "step": 606
+    },
+    {
+      "epoch": 1.8173652694610778,
+      "grad_norm": 0.4062684446179707,
+      "learning_rate": 2.043458732777831e-07,
+      "loss": 0.2365,
+      "step": 607
+    },
+    {
+      "epoch": 1.8203592814371259,
+      "grad_norm": 0.4816795985463741,
+      "learning_rate": 1.9774508203096843e-07,
+      "loss": 0.2794,
+      "step": 608
+    },
+    {
+      "epoch": 1.8233532934131738,
+      "grad_norm": 0.4142237332912122,
+      "learning_rate": 1.9125050709908388e-07,
+      "loss": 0.247,
+      "step": 609
+    },
+    {
+      "epoch": 1.8263473053892216,
+      "grad_norm": 0.45111327376712257,
+      "learning_rate": 1.8486229212919482e-07,
+      "loss": 0.2791,
+      "step": 610
+    },
+    {
+      "epoch": 1.8293413173652695,
+      "grad_norm": 0.4045387611084571,
+      "learning_rate": 1.7858057841589281e-07,
+      "loss": 0.2593,
+      "step": 611
+    },
+    {
+      "epoch": 1.8323353293413174,
+      "grad_norm": 0.48833180119294667,
+      "learning_rate": 1.7240550489817652e-07,
+      "loss": 0.2794,
+      "step": 612
+    },
+    {
+      "epoch": 1.8353293413173652,
+      "grad_norm": 0.5134652341553744,
+      "learning_rate": 1.66337208156373e-07,
+      "loss": 0.258,
+      "step": 613
+    },
+    {
+      "epoch": 1.8383233532934131,
+      "grad_norm": 0.4580098802599602,
+      "learning_rate": 1.6037582240912175e-07,
+      "loss": 0.278,
+      "step": 614
+    },
+    {
+      "epoch": 1.841317365269461,
+      "grad_norm": 0.42572130384500223,
+      "learning_rate": 1.5452147951040165e-07,
+      "loss": 0.2735,
+      "step": 615
+    },
+    {
+      "epoch": 1.8443113772455089,
+      "grad_norm": 0.49952287085840236,
+      "learning_rate": 1.4877430894662037e-07,
+      "loss": 0.3055,
+      "step": 616
+    },
+    {
+      "epoch": 1.8473053892215567,
+      "grad_norm": 0.4456446654322269,
+      "learning_rate": 1.4313443783374405e-07,
+      "loss": 0.2502,
+      "step": 617
+    },
+    {
+      "epoch": 1.8502994011976048,
+      "grad_norm": 0.39652383268972746,
+      "learning_rate": 1.3760199091449045e-07,
+      "loss": 0.2297,
+      "step": 618
+    },
+    {
+      "epoch": 1.8532934131736527,
+      "grad_norm": 0.38116740406360927,
+      "learning_rate": 1.3217709055556638e-07,
+      "loss": 0.224,
+      "step": 619
+    },
+    {
+      "epoch": 1.8562874251497006,
+      "grad_norm": 0.40926608753281013,
+      "learning_rate": 1.268598567449647e-07,
+      "loss": 0.2689,
+      "step": 620
+    },
+    {
+      "epoch": 1.8592814371257484,
+      "grad_norm": 0.4872409527177046,
+      "learning_rate": 1.2165040708930763e-07,
+      "loss": 0.2612,
+      "step": 621
+    },
+    {
+      "epoch": 1.8622754491017965,
+      "grad_norm": 0.3708682816190577,
+      "learning_rate": 1.1654885681124661e-07,
+      "loss": 0.251,
+      "step": 622
+    },
+    {
+      "epoch": 1.8652694610778444,
+      "grad_norm": 0.44647910236931404,
+      "learning_rate": 1.1155531874691372e-07,
+      "loss": 0.2895,
+      "step": 623
+    },
+    {
+      "epoch": 1.8682634730538923,
+      "grad_norm": 0.42922686622461886,
+      "learning_rate": 1.0666990334342708e-07,
+      "loss": 0.2407,
+      "step": 624
+    },
+    {
+      "epoch": 1.8712574850299402,
+      "grad_norm": 0.41820624022834957,
+      "learning_rate": 1.0189271865644445e-07,
+      "loss": 0.2591,
+      "step": 625
+    },
+    {
+      "epoch": 1.874251497005988,
+      "grad_norm": 0.5549232833754368,
+      "learning_rate": 9.722387034777847e-08,
+      "loss": 0.3214,
+      "step": 626
+    },
+    {
+      "epoch": 1.877245508982036,
+      "grad_norm": 0.4645509555882806,
+      "learning_rate": 9.266346168305518e-08,
+      "loss": 0.3159,
+      "step": 627
+    },
+    {
+      "epoch": 1.8802395209580838,
+      "grad_norm": 0.39857357188374193,
+      "learning_rate": 8.821159352943142e-08,
+      "loss": 0.26,
+      "step": 628
+    },
+    {
+      "epoch": 1.8832335329341316,
+      "grad_norm": 0.4460879124858831,
+      "learning_rate": 8.38683643533661e-08,
+      "loss": 0.2694,
+      "step": 629
+    },
+    {
+      "epoch": 1.8862275449101795,
+      "grad_norm": 0.46790740896105454,
+      "learning_rate": 7.963387021843683e-08,
+      "loss": 0.2726,
+      "step": 630
+    },
+    {
+      "epoch": 1.8892215568862274,
+      "grad_norm": 0.45465547157053104,
+      "learning_rate": 7.550820478322285e-08,
+      "loss": 0.2995,
+      "step": 631
+    },
+    {
+      "epoch": 1.8922155688622755,
+      "grad_norm": 0.41463122040688744,
+      "learning_rate": 7.149145929922607e-08,
+      "loss": 0.2283,
+      "step": 632
+    },
+    {
+      "epoch": 1.8952095808383234,
+      "grad_norm": 0.444658524129169,
+      "learning_rate": 6.758372260885714e-08,
+      "loss": 0.2793,
+      "step": 633
+    },
+    {
+      "epoch": 1.8982035928143712,
+      "grad_norm": 0.49252922982085995,
+      "learning_rate": 6.378508114346982e-08,
+      "loss": 0.2588,
+      "step": 634
+    },
+    {
+      "epoch": 1.9011976047904193,
+      "grad_norm": 0.5097029065302707,
+      "learning_rate": 6.009561892144744e-08,
+      "loss": 0.252,
+      "step": 635
+    },
+    {
+      "epoch": 1.9041916167664672,
+      "grad_norm": 0.4600102382713019,
+      "learning_rate": 5.651541754634726e-08,
+      "loss": 0.2439,
+      "step": 636
+    },
+    {
+      "epoch": 1.907185628742515,
+      "grad_norm": 0.4731442067701136,
+      "learning_rate": 5.304455620509297e-08,
+      "loss": 0.2792,
+      "step": 637
+    },
+    {
+      "epoch": 1.910179640718563,
+      "grad_norm": 0.4584527434850442,
+      "learning_rate": 4.968311166622553e-08,
+      "loss": 0.267,
+      "step": 638
+    },
+    {
+      "epoch": 1.9131736526946108,
+      "grad_norm": 0.48435725016827297,
+      "learning_rate": 4.643115827820399e-08,
+      "loss": 0.2391,
+      "step": 639
+    },
+    {
+      "epoch": 1.9161676646706587,
+      "grad_norm": 0.49426219440395347,
+      "learning_rate": 4.328876796776071e-08,
+      "loss": 0.307,
+      "step": 640
+    },
+    {
+      "epoch": 1.9191616766467066,
+      "grad_norm": 0.45510609076638586,
+      "learning_rate": 4.0256010238310936e-08,
+      "loss": 0.2505,
+      "step": 641
+    },
+    {
+      "epoch": 1.9221556886227544,
+      "grad_norm": 0.43120296001203884,
+      "learning_rate": 3.733295216841626e-08,
+      "loss": 0.2824,
+      "step": 642
+    },
+    {
+      "epoch": 1.9251497005988023,
+      "grad_norm": 0.4476162932816316,
+      "learning_rate": 3.451965841029914e-08,
+      "loss": 0.2631,
+      "step": 643
+    },
+    {
+      "epoch": 1.9281437125748502,
+      "grad_norm": 0.403496770287237,
+      "learning_rate": 3.181619118841517e-08,
+      "loss": 0.3009,
+      "step": 644
+    },
+    {
+      "epoch": 1.931137724550898,
+      "grad_norm": 0.5068605554399377,
+      "learning_rate": 2.9222610298074717e-08,
+      "loss": 0.2515,
+      "step": 645
+    },
+    {
+      "epoch": 1.9341317365269461,
+      "grad_norm": 0.44106003442980724,
+      "learning_rate": 2.673897310412288e-08,
+      "loss": 0.2945,
+      "step": 646
+    },
+    {
+      "epoch": 1.937125748502994,
+      "grad_norm": 0.37135313943250803,
+      "learning_rate": 2.4365334539667717e-08,
+      "loss": 0.2316,
+      "step": 647
+    },
+    {
+      "epoch": 1.9401197604790419,
+      "grad_norm": 0.44980623901916356,
+      "learning_rate": 2.210174710486679e-08,
+      "loss": 0.2603,
+      "step": 648
+    },
+    {
+      "epoch": 1.94311377245509,
+      "grad_norm": 0.4124149568187268,
+      "learning_rate": 1.99482608657664e-08,
+      "loss": 0.2523,
+      "step": 649
+    },
+    {
+      "epoch": 1.9461077844311379,
+      "grad_norm": 0.4701293082440937,
+      "learning_rate": 1.7904923453193056e-08,
+      "loss": 0.2958,
+      "step": 650
+    },
+    {
+      "epoch": 1.9491017964071857,
+      "grad_norm": 0.4298668251857666,
+      "learning_rate": 1.5971780061701524e-08,
+      "loss": 0.2558,
+      "step": 651
+    },
+    {
+      "epoch": 1.9520958083832336,
+      "grad_norm": 0.43862178397593465,
+      "learning_rate": 1.4148873448573408e-08,
+      "loss": 0.2391,
+      "step": 652
+    },
+    {
+      "epoch": 1.9550898203592815,
+      "grad_norm": 0.39468990600361525,
+      "learning_rate": 1.2436243932872349e-08,
+      "loss": 0.2689,
+      "step": 653
+    },
+    {
+      "epoch": 1.9580838323353293,
+      "grad_norm": 0.43262578248306477,
+      "learning_rate": 1.0833929394552523e-08,
+      "loss": 0.2307,
+      "step": 654
+    },
+    {
+      "epoch": 1.9610778443113772,
+      "grad_norm": 0.4136424529236092,
+      "learning_rate": 9.341965273621522e-09,
+      "loss": 0.2822,
+      "step": 655
+    },
+    {
+      "epoch": 1.964071856287425,
+      "grad_norm": 0.44849445179422825,
+      "learning_rate": 7.96038456935322e-09,
+      "loss": 0.284,
+      "step": 656
+    },
+    {
+      "epoch": 1.967065868263473,
+      "grad_norm": 0.44510919927885495,
+      "learning_rate": 6.6892178395611125e-09,
+      "loss": 0.3037,
+      "step": 657
+    },
+    {
+      "epoch": 1.9700598802395208,
+      "grad_norm": 0.5166900752958025,
+      "learning_rate": 5.528493199922769e-09,
+      "loss": 0.3056,
+      "step": 658
+    },
+    {
+      "epoch": 1.9730538922155687,
+      "grad_norm": 0.43276198169898283,
+      "learning_rate": 4.478236323355312e-09,
+      "loss": 0.2415,
+      "step": 659
+    },
+    {
+      "epoch": 1.9760479041916168,
+      "grad_norm": 0.4688173279433659,
+      "learning_rate": 3.538470439448105e-09,
+      "loss": 0.3194,
+      "step": 660
+    },
+    {
+      "epoch": 1.9790419161676647,
+      "grad_norm": 0.4267989111079099,
+      "learning_rate": 2.709216333952602e-09,
+      "loss": 0.2815,
+      "step": 661
+    },
+    {
+      "epoch": 1.9820359281437125,
+      "grad_norm": 0.4061632736115782,
+      "learning_rate": 1.9904923483171632e-09,
+      "loss": 0.2276,
+      "step": 662
+    },
+    {
+      "epoch": 1.9850299401197606,
+      "grad_norm": 0.45208406420882724,
+      "learning_rate": 1.3823143792851545e-09,
+      "loss": 0.3126,
+      "step": 663
+    },
+    {
+      "epoch": 1.9880239520958085,
+      "grad_norm": 0.4886730565980843,
+      "learning_rate": 8.846958785418969e-10,
+      "loss": 0.3081,
+      "step": 664
+    },
+    {
+      "epoch": 1.9910179640718564,
+      "grad_norm": 0.4534649044675671,
+      "learning_rate": 4.97647852417682e-10,
+      "loss": 0.2458,
+      "step": 665
+    },
+    {
+      "epoch": 1.9940119760479043,
+      "grad_norm": 0.41481136000015817,
+      "learning_rate": 2.2117886164407797e-10,
+      "loss": 0.2527,
+      "step": 666
+    },
+    {
+      "epoch": 1.9970059880239521,
+      "grad_norm": 0.3757913605076721,
+      "learning_rate": 5.529502116519148e-11,
+      "loss": 0.2317,
+      "step": 667
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.43418259959714856,
+      "learning_rate": 0.0,
+      "loss": 0.1991,
+      "step": 668
+    },
+    {
+      "epoch": 2.0,
+      "step": 668,
+      "total_flos": 21199392964608.0,
+      "train_loss": 0.31201071171703454,
+      "train_runtime": 902.9926,
+      "train_samples_per_second": 11.803,
+      "train_steps_per_second": 0.74
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 668,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 5000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 21199392964608.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}